Introspection.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Introspection.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "exQzRKEpMaVQ",
        "dd0hyp2_MaZb",
        "fmSIB_Dufl9j"
      ],
      "authorship_tag": "ABX9TyO2p9d989RDpwa+1IKkLJNK",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/surajsrivathsa/ovgu_deeplearning/blob/master/Introspection.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UNGZH5mPFjbf",
        "colab_type": "text"
      },
      "source": [
        "**Project Assignment 4:** Introspection\n",
        "\n",
        "**Team members:**\n",
        "1. Sanjeeth Busnur Indushekar: 224133 : sanjeeth.busnur@st.ovgu.de\n",
        "2. Suraj Shashidhar: 230052 : suraj.shashidhar@st.ovgu.de"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FcIEgqJpG3kH",
        "colab_type": "text"
      },
      "source": [
        "Tasks To be done\n",
        "\n",
        "4.1 gradient-based saliency map (sensitivity analysis)\n",
        "\n",
        "- - Run a batch of inputs through the trained model.\n",
        "\n",
        "Wrap this in a GradientTape where you watch the input batch.\n",
        "(batch size can be 1 if you just like to produce a single saliency map)\n",
        "Compute the gradient for a particular logit or its softmax output with respect to the input.\n",
        "\n",
        "- This tells us how a change in each input pixel would affect the class output.\n",
        "\n",
        "- This already gives you a batch of gradient-based saliency maps!\n",
        "\n",
        "- Plot the saliency map next to the original image or superimpose it.\n",
        "\n",
        "- Do the saliency maps seem to make sense? How would you interpret them?\n",
        "\n",
        "Hint: It makes sense to take the sign of the gradient into account when interpreting them.\n",
        "\n",
        "Negative gradients indicate a decrease in output value, positive gradients an increase.\n",
        "\n",
        "4.2 Activation Maximization\n",
        "\n",
        "Extend the code from 4.1 to create an optimal input for a particular class.\n",
        "\n",
        "- Add the gradients (i.e. the saliency map) to the input. You should multiply the gradients with a small constant (like a learning rate) to make sure the inputs aren’t changed too much.\n",
        "\n",
        "- Do multiple update steps to make the input more optimal for the chosen output class.\n",
        "(I recommend implementing your own update step, but you are allowed to use a keras training loop to optimize the input.)\n",
        "\n",
        "- Does the resulting input look natural?\n",
        "\n",
        "- How does the inputs change when applying many steps of optimization?\n",
        "\n",
        "- How do the optimal inputs differ when initializing the optimization with random noise instead of real examples?\n",
        "\n",
        "Bonus: Apply regularization strategies to make the optimal input more natural-looking.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Tm-7nvSF4Jbd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import os\n",
        "%tensorflow_version 2.x\n",
        "import tensorflow as tf\n",
        "from sklearn import preprocessing\n",
        "from google.colab import files\n",
        "from google.colab import drive\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "import copy\n",
        "import numpy as np\n",
        "from datasets import MNISTDataset"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tJzT37jxLvHH",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.preprocessing import image\n",
        "from tensorflow.keras import datasets, layers, models\n",
        "from tensorflow.keras import initializers\n",
        "import tensorboard\n",
        "import time\n",
        "from datetime import datetime\n",
        "from keras import backend as K"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wQ2-hrUA4qdc",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "bca1899b-c948-4e72-db13-1df0dec587f4"
      },
      "source": [
        "print('tensorflow {}'.format(tf.__version__))\n",
        "print(\"keras {}\".format(keras.__version__))"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tensorflow 2.3.0\n",
            "keras 2.4.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hW-aKSx5MTHc",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 316
        },
        "outputId": "841c5084-df37-4d56-f830-52a4949ce657"
      },
      "source": [
        "mnist = tf.keras.datasets.mnist\n",
        "(train_images, train_labels), (test_images, test_labels) = mnist.load_data()\n",
        "plt.imshow(train_images[4], cmap=\"Greys_r\")\n",
        "print(train_labels[1:10])"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n",
            "11493376/11490434 [==============================] - 0s 0us/step\n",
            "[0 4 1 9 2 1 3 1 4]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uRR5LQ-0MTKK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "d68cf43e-aa51-4f8c-9d57-e680f2cd28e9"
      },
      "source": [
        "data = MNISTDataset(train_images.reshape([-1, 784]), train_labels, \n",
        "                    test_images.reshape([-1, 784]), test_labels,\n",
        "                    batch_size=128)\n",
        "print(train_images.reshape([-1, 784]).shape)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(60000, 784)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0l0lWO8bMTQX",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "72e48cfb-7773-4ec8-ac70-09bd31e82ce2"
      },
      "source": [
        "print(train_images.shape)\n",
        "print(train_labels.shape)\n",
        "print(test_images.shape)\n",
        "print(test_labels.shape)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(60000, 28, 28)\n",
            "(60000,)\n",
            "(10000, 28, 28)\n",
            "(10000,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_S2qT04PTNsj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "ecd487dd-429d-4a97-8121-911aaf1eda4f"
      },
      "source": [
        "new_train_images = tf.cast(tf.expand_dims(train_images, axis = -1),dtype=\"float32\")\n",
        "print(new_train_images.shape)\n",
        "\n",
        "new_test_images = tf.cast(tf.expand_dims(test_images, axis=-1 ),dtype=\"float32\")\n",
        "print(new_test_images.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(60000, 28, 28, 1)\n",
            "(10000, 28, 28, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "25bDbj07nu_u",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "introspection_training_images = tf.cast(tf.expand_dims(train_images, axis = -1),dtype=\"float32\")\n",
        "introspection_testing_images = tf.cast(tf.expand_dims(test_images, axis=-1 ),dtype=\"float32\")\n",
        "\n",
        "introspection_training_labels = tf.convert_to_tensor(train_labels, dtype=tf.float32)\n",
        "introspection_testing_labels = tf.convert_to_tensor(test_labels, dtype=tf.float32)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "exQzRKEpMaVQ",
        "colab_type": "text"
      },
      "source": [
        "# Pretrain a simple CNN model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "819hgv8_MnGS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def create_cnn_model():\n",
        "  model = keras.Sequential()\n",
        "  model.add(layers.Conv2D(filters=8,kernel_size=(3,3),padding=\"same\",activation=\"relu\" , input_shape=(28, 28, 1), ))\n",
        "  model.add(layers.BatchNormalization())\n",
        "  model.add(layers.Conv2D(filters=16,kernel_size=(3,3),padding=\"same\",activation=\"relu\" ))\n",
        "  model.add(layers.BatchNormalization())\n",
        "  model.add(layers.MaxPool2D(pool_size=(2,2), strides=2))\n",
        "  model.add(layers.Conv2D(filters=32,kernel_size=(3,3),padding=\"same\", activation=\"relu\"))\n",
        "  model.add(layers.BatchNormalization())\n",
        "  model.add(layers.Conv2D(filters=64,kernel_size=(3,3),padding=\"same\", activation=\"relu\"))\n",
        "  model.add(layers.BatchNormalization())\n",
        "  model.add(layers.MaxPool2D(pool_size=(2,2), strides=2))\n",
        "  model.add(layers.Conv2D(filters=128,kernel_size=(3,3),padding=\"same\", activation=\"relu\"))\n",
        "  model.add(layers.BatchNormalization())\n",
        "  model.add(layers.Conv2D(filters=256,kernel_size=(3,3),padding=\"same\", activation=\"relu\"))\n",
        "  model.add(layers.BatchNormalization())\n",
        "  model.add(layers.MaxPool2D(pool_size=(2,2), strides=2))\n",
        "  model.add(layers.Conv2D(filters=512,kernel_size=(3,3),padding=\"same\", activation=\"relu\"))\n",
        "  model.add(layers.BatchNormalization())\n",
        "  model.add(layers.Conv2D(filters=64,kernel_size=(1,1),padding=\"same\", activation=\"relu\"))\n",
        "  model.add(layers.BatchNormalization())\n",
        "  model.add(layers.Flatten(input_shape=(3,3,64)))\n",
        "  model.add(layers.Dense(units=256, activation=\"relu\"))\n",
        "  model.add(layers.Dense(units=10,activation=\"softmax\"))\n",
        "  model.summary()\n",
        "  return model;"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LDMdOqWeMnMC",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 941
        },
        "outputId": "c6ec18ea-363e-414f-c65b-1f007c5aa4bd"
      },
      "source": [
        "vgg_cnn_model = create_cnn_model()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 28, 28, 8)         80        \n",
            "_________________________________________________________________\n",
            "batch_normalization (BatchNo (None, 28, 28, 8)         32        \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 28, 28, 16)        1168      \n",
            "_________________________________________________________________\n",
            "batch_normalization_1 (Batch (None, 28, 28, 16)        64        \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 14, 14, 16)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 14, 14, 32)        4640      \n",
            "_________________________________________________________________\n",
            "batch_normalization_2 (Batch (None, 14, 14, 32)        128       \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 14, 14, 64)        18496     \n",
            "_________________________________________________________________\n",
            "batch_normalization_3 (Batch (None, 14, 14, 64)        256       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 7, 7, 64)          0         \n",
            "_________________________________________________________________\n",
            "conv2d_4 (Conv2D)            (None, 7, 7, 128)         73856     \n",
            "_________________________________________________________________\n",
            "batch_normalization_4 (Batch (None, 7, 7, 128)         512       \n",
            "_________________________________________________________________\n",
            "conv2d_5 (Conv2D)            (None, 7, 7, 256)         295168    \n",
            "_________________________________________________________________\n",
            "batch_normalization_5 (Batch (None, 7, 7, 256)         1024      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 3, 3, 256)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_6 (Conv2D)            (None, 3, 3, 512)         1180160   \n",
            "_________________________________________________________________\n",
            "batch_normalization_6 (Batch (None, 3, 3, 512)         2048      \n",
            "_________________________________________________________________\n",
            "conv2d_7 (Conv2D)            (None, 3, 3, 64)          32832     \n",
            "_________________________________________________________________\n",
            "batch_normalization_7 (Batch (None, 3, 3, 64)          256       \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 576)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 256)               147712    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 10)                2570      \n",
            "=================================================================\n",
            "Total params: 1,761,002\n",
            "Trainable params: 1,758,842\n",
            "Non-trainable params: 2,160\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OcCP4CupMnJX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vgg_cnn_model.compile(optimizer=\"adam\", loss = tf.keras.losses.sparse_categorical_crossentropy, metrics=[\"accuracy\"])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EKrKfFvjSs-7",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 141
        },
        "outputId": "3de9058e-91ca-441d-de51-a37771c67fc5"
      },
      "source": [
        "vgg_cnn_model.fit(new_train_images, train_labels,  epochs=3, batch_size=128, validation_data= (new_test_images, test_labels))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/3\n",
            "469/469 [==============================] - 5s 11ms/step - loss: 0.1188 - accuracy: 0.9636 - val_loss: 0.0642 - val_accuracy: 0.9797\n",
            "Epoch 2/3\n",
            "469/469 [==============================] - 5s 10ms/step - loss: 0.0348 - accuracy: 0.9891 - val_loss: 0.0317 - val_accuracy: 0.9904\n",
            "Epoch 3/3\n",
            "469/469 [==============================] - 5s 10ms/step - loss: 0.0269 - accuracy: 0.9919 - val_loss: 0.0397 - val_accuracy: 0.9884\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f249e7a7390>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 103
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CNqID72rStCD",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "16b8b6fc-def2-408d-9d5e-3ad2916b5f26"
      },
      "source": [
        "vgg_cnn_model.save(\"/content/vgg_cnn_model_introspection\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: /content/vgg_cnn_model_introspection/assets\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qhi1ZRboyGsF",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 141
        },
        "outputId": "d5b6adc2-0983-4136-f299-db37279a57ea"
      },
      "source": [
        "!zip -r vgg_cnn_model_introspection.zip /content/vgg_cnn_model_introspection"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  adding: content/vgg_cnn_model_introspection/ (stored 0%)\n",
            "  adding: content/vgg_cnn_model_introspection/saved_model.pb (deflated 91%)\n",
            "  adding: content/vgg_cnn_model_introspection/variables/ (stored 0%)\n",
            "  adding: content/vgg_cnn_model_introspection/variables/variables.data-00001-of-00002 (deflated 8%)\n",
            "  adding: content/vgg_cnn_model_introspection/variables/variables.index (deflated 75%)\n",
            "  adding: content/vgg_cnn_model_introspection/variables/variables.data-00000-of-00002 (deflated 83%)\n",
            "  adding: content/vgg_cnn_model_introspection/assets/ (stored 0%)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nsQayqwxyGwE",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dd0hyp2_MaZb",
        "colab_type": "text"
      },
      "source": [
        "# Saliency Maps\n",
        "\n",
        "This is the side by side comparision of images with mask of number one applied.\n",
        "\n",
        "**Process**\n",
        "\n",
        "- Predict the softmax output of the model.\n",
        "\n",
        "- Apply mask on softmax/logits output. For example if we want to find the output number ones contribution or sensitivity then except number one softmax make everything zero, same way do for other numbers separately. Hence we end up with 10 masks.\n",
        "\n",
        "- Apply these masks on softmax output and find the losses. Obviously losses will be high for most of the numbers as we have masked the relvenat part of softmax out\n",
        "\n",
        "- Backpropagate and now find the derivative of masked softmax wrt input image, Since all softmax output positions except one position is zero gradient can only flow back through this position.\n",
        "\n",
        "- This is not guided backprop as we are not applying subsequent relu masks on intermeditate layers featuremaps but only on output.\n",
        "\n",
        "- At the end get the gradients and plot them side by side wrt to input image\n",
        "\n",
        "**Sensitivity analysis for number 1**\n",
        "\n",
        "- We can see that numbers **6, 8, 0** which doesn't have stick shape has no bearing or contribution of softmax(even a little softmax of one is not present in these numbers). S**o these numbers are insensitive to the number one.**\n",
        "\n",
        "- For some numbers like **3, 9 and 4** we can see a slight presence of one when the numbers are slanted. **Less sensitivity**\n",
        "\n",
        "- For number **7** the presence of softmax ones contribution can be clearly seen especially at last image. **Some sensitivity.**\n",
        "\n",
        "- **For contribution of one to one itself, we can see some strong sensitivity.** But thing to note here is for the second one(the one in 3rd row which is really straight and not slant) **it is surprising to see that one is getting recognized due to sensitivity of irrelvenat pixels rather the stick like figure, for one that are slant the activations are from stick.** **Same pattern is present in 7 too if we can see at the last 7 where slnated 7 is getting recognized.**\n",
        "\n",
        "- **This clearly demonstrates that CNN based detection models are non-human like, even though we think them to be. Even after 50 - 60k examples CNN was not able to learn the context better than humans where humans have lower resolution than machines. This is a clear demonstration that we need Better models than CNNs in the future**.\n",
        "\n",
        "**Sensitivity analysis for number 8**\n",
        "\n",
        "- Surprisingly number 1 and 7 which are least related to number 8 are getting activated. Seems like slanted stick like pattern is getting activated everytime.\n",
        "\n",
        "- Numbers 9 are getting activated which are no surprise as upper part of 9 looks like a half of 8. But 9 was activated mainly due to its lower part and garbage pixels. which raises doubts on ability of CNN to understand semantic meaning of images.\n",
        "\n",
        "- 3 should have been activated as it looks like a half of 8, but it dod not happen."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pRQ7FCvieioU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 158
        },
        "outputId": "a8baa0f0-6eb6-4423-9d1c-905f684a5962"
      },
      "source": [
        "!unzip \"/content/vgg_cnn_model_introspection.zip\" "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Archive:  /content/vgg_cnn_model_introspection.zip\n",
            "   creating: content/vgg_cnn_model_introspection/\n",
            "  inflating: content/vgg_cnn_model_introspection/saved_model.pb  \n",
            "   creating: content/vgg_cnn_model_introspection/variables/\n",
            "  inflating: content/vgg_cnn_model_introspection/variables/variables.data-00001-of-00002  \n",
            "  inflating: content/vgg_cnn_model_introspection/variables/variables.index  \n",
            "  inflating: content/vgg_cnn_model_introspection/variables/variables.data-00000-of-00002  \n",
            "   creating: content/vgg_cnn_model_introspection/assets/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-8MyMib4evMg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vgg_cnn_model = keras.models.load_model(filepath=\"/content/content/vgg_cnn_model_introspection\")\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k2BVHCAdfFhS",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 941
        },
        "outputId": "47d28550-6479-4da7-c437-1c26504a5881"
      },
      "source": [
        "vgg_cnn_model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 28, 28, 8)         80        \n",
            "_________________________________________________________________\n",
            "batch_normalization (BatchNo (None, 28, 28, 8)         32        \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 28, 28, 16)        1168      \n",
            "_________________________________________________________________\n",
            "batch_normalization_1 (Batch (None, 28, 28, 16)        64        \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 14, 14, 16)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 14, 14, 32)        4640      \n",
            "_________________________________________________________________\n",
            "batch_normalization_2 (Batch (None, 14, 14, 32)        128       \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 14, 14, 64)        18496     \n",
            "_________________________________________________________________\n",
            "batch_normalization_3 (Batch (None, 14, 14, 64)        256       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 7, 7, 64)          0         \n",
            "_________________________________________________________________\n",
            "conv2d_4 (Conv2D)            (None, 7, 7, 128)         73856     \n",
            "_________________________________________________________________\n",
            "batch_normalization_4 (Batch (None, 7, 7, 128)         512       \n",
            "_________________________________________________________________\n",
            "conv2d_5 (Conv2D)            (None, 7, 7, 256)         295168    \n",
            "_________________________________________________________________\n",
            "batch_normalization_5 (Batch (None, 7, 7, 256)         1024      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 3, 3, 256)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_6 (Conv2D)            (None, 3, 3, 512)         1180160   \n",
            "_________________________________________________________________\n",
            "batch_normalization_6 (Batch (None, 3, 3, 512)         2048      \n",
            "_________________________________________________________________\n",
            "conv2d_7 (Conv2D)            (None, 3, 3, 64)          32832     \n",
            "_________________________________________________________________\n",
            "batch_normalization_7 (Batch (None, 3, 3, 64)          256       \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 576)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 256)               147712    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 10)                2570      \n",
            "=================================================================\n",
            "Total params: 1,761,002\n",
            "Trainable params: 1,758,842\n",
            "Non-trainable params: 2,160\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8vdf0iOgMTOU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "572275fe-eda7-49a9-f077-8733682dec4a"
      },
      "source": [
        "introspection_image = introspection_testing_images[0, :, :, :]\n",
        "plt.imshow(introspection_image[:, :, 0], cmap=\"binary\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fc30012bc88>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oCrh0JPkH3Tl",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "5b7f8b82-4dbb-40a5-e33b-eef7aa161d0a"
      },
      "source": [
        "introspection_prediction = vgg_cnn_model.predict(introspection_testing_images[0:8, :, :, :])\n",
        "introspection_prediction.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(8, 10)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dUPiezuj4YBx",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "96c62155-5455-4186-f3f7-ecfffe273e0a"
      },
      "source": [
        "plt.imshow(introspection_testing_images[7,:,:,0], cmap=\"binary\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fbddff894a8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CQBtBmspH3Rf",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 70
        },
        "outputId": "746e0e33-dac2-41ac-ef9b-272fad414a23"
      },
      "source": [
        "introspection_prediction[7]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([7.5519985e-08, 2.8737236e-06, 5.9842296e-07, 2.4699690e-07,\n",
              "       4.1525485e-03, 2.5558572e-06, 4.9045508e-07, 1.7699102e-07,\n",
              "       8.8364504e-06, 9.9583167e-01], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aCHdBqxsH3Ph",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "11087fcc-e818-42c7-844b-57c3c8d74e6e"
      },
      "source": [
        "plt.imshow(introspection_testing_images[7, :, :, 0], cmap=\"binary\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fbe01f9b6d8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r6AeBpM76BM0",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "zero_mask = tf.tile(tf.Variable(initial_value=tf.constant([1,0,0,0,0,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "one_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,1,0,0,0,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "two_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,1,0,0,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "three_mask = tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,1,0,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "four_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,1,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "five_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,1,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "six_mask = tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,0,1,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "seven_mask = tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,0,0,1,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "eight_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,0,0,0,1,0], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "nine_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,0,0,0,0,1], dtype=\"float32\", shape=(1,10))) , multiples=[10000,1])\n",
        "\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MajrUJH46BFC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "\n",
        "def record_tape(introspection_testing_images, introspection_testing_labels, mask_list=[zero_mask]):\n",
        "  with tf.GradientTape() as tape:\n",
        "    tape.watch(introspection_testing_labels)\n",
        "    tape.watch(introspection_testing_images)\n",
        "    introspection_prediction = vgg_cnn_model(introspection_testing_images)\n",
        "    print(introspection_prediction.shape)\n",
        "    masked_output = []\n",
        "    masked_loss = []\n",
        "\n",
        "    # We need to find the loss only wrt one logits, hence set all the other logits output to zero by multiplying with binary mask\n",
        "    for msk in mask_list:\n",
        "      masked_output.append(tf.multiply(msk, introspection_prediction))\n",
        "\n",
        "    #zero_position_output = tf.multiply(zero_mask,introspection_prediction)\n",
        "    #ones_position_output = tf.multiply(one_mask,introspection_prediction)\n",
        "    \n",
        "    #Not used anywhere, just there for simple cross check\n",
        "    complete_loss = tf.keras.losses.sparse_categorical_crossentropy(introspection_testing_labels, introspection_prediction)\n",
        "    \n",
        "\n",
        "    #masked loss corresponding to the each masked output\n",
        "    \"\"\"\n",
        "    For example: if picture is of number 5 and mask of 5 is applied then every position except 5 is made zero so gradient doesn't flow back.\n",
        "    , for mask of any other number we have higher losss as it doesn't correspond to 5.\n",
        "    Loss will be high for other numbers than picture\n",
        "    \"\"\"\n",
        "    for msk_op in masked_output:\n",
        "      masked_loss.append(tf.keras.losses.sparse_categorical_crossentropy(introspection_testing_labels, msk_op))\n",
        "\n",
        "    #ones_grad_loss = tf.keras.losses.sparse_categorical_crossentropy(test_labels,ones_position_output)\n",
        "\n",
        "  return [tape, masked_loss];\n",
        "\n",
        "#grads = tape.gradient(zero_grad_loss,introspection_testing_images)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kunOBN0urNMG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "603ad6d4-e18f-4baf-f5c8-a9bb6e93f8ab"
      },
      "source": [
        "tape, masked_loss = record_tape(introspection_testing_images, introspection_testing_labels, mask_list=[one_mask])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(10000, 10)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XsK6ItIEkjqa",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def calculate_masked_grads(introspection_testing_images, introspection_testing_labels, tape, masked_loss):\n",
        "  masked_grads = []\n",
        "\n",
        "  for msk_loss in masked_loss:\n",
        "    masked_grads.append(tape.gradient(msk_loss,introspection_testing_images))\n",
        "  \n",
        "  return masked_grads;"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_AHlmi7Hkvbw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "masked_grads = calculate_masked_grads(introspection_testing_images, introspection_testing_labels, tape, masked_loss)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BsbT7NtAuIND",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "88ae4cca-e95e-4ba2-d24d-1a44090d31c3"
      },
      "source": [
        "masked_grads[0].shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "TensorShape([10000, 28, 28, 1])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UXYIPTimxchN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "masked_grads[0][2,:,:,:]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ckr5WiSGxEYt",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "93f0dd26-63cd-48d7-8754-c8c20cc77475"
      },
      "source": [
        "plt.imshow(tf.nn.relu(masked_grads[0][2,:,:,0]), cmap = \"binary\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fc30ce37cf8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAU0UlEQVR4nO3da2yVZbYH8P+yXFpKAUuh1FKpIBAIl864423QjDM6KJHoGDXjhwnHTGSMl8zESTxGQ0Y/aU7OzGQSj6MoZpiJx3Gig5doVCQmxERHt1guglwtl1JaCuVSKBRwnQ99MVX6rlX3u/d+t+f5/xLSdv/7dD9sutjtXu/zPKKqIKL//85LewJEVBwsdqJAsNiJAsFiJwoEi50oEEOKeWc1NTXa2NhYzLv8Xvjqq6/M/Lzz7P+TrY6KiOQ0p3yx/m7e3+vMmTNmXlZWlvN9e49Lmo+b1yGz5tbS0oLOzs4BPyFRsYvI9QD+DKAMwHOq+oT1+Y2NjchmsznfXyl/U1u8b9oTJ06YeWVlpZn39vbGZsOGDTPHFlpPT09sVlFRYY49fPiwmY8ePdrMu7u7YzPvcRk6dKiZe99vSQr29OnT5tghQ+LLNpPJxGY5/xgvImUA/gfADQBmArhDRGbm+vWIqLCS/M5+KYBtqrpDVXsB/APATfmZFhHlW5Jirwewu9/He6LbvkFEFotIVkSy+/fvT3B3RJREwV+NV9WlqppR1cy4ceMKfXdEFCNJsbcCaOj38cToNiIqQUmK/RMAU0XkIhEZBuAXAF7Pz7SIKN9ybr2p6mkRuQ/AO+hrvT2vqp87Y8y2gteusNohnZ2d5tiamhozt1pEXj5q1ChzrNUqAYCjR4+audd6O3jwYGw2YcIEc2yhlZeX5zz25MmTie7bGj9y5EhzbHt7u5nX1taaudX2A4CqqqrY7Pjx4+ZY7/stTqI+u6q+BeCtJF+DiIqDl8sSBYLFThQIFjtRIFjsRIFgsRMFgsVOFIiirmcXEXMNs7eu21oq2tXVZY6trq4281WrVpn5jTfeGJt5SzGHDx9u5rn2Tc8qZC/9wIEDZu71q61rAOrq6syx3hJYbyno2LFjYzNvnYbXR/e+V60+OmAva/YeU2/JdBw+sxMFgsVOFAgWO1EgWOxEgWCxEwWCxU4UiKK23gB7+2BviWuSQyj37t1r5lZrDQC2b98em02ZMsUcu3v3bjP32jQjRowwc6u95bUcvbah1b4aDKu95rXOvB1cvSWwu3btis0mT55sjt20aZOZz5gxw8xPnTpl5tbfPcmyYAuf2YkCwWInCgSLnSgQLHaiQLDYiQLBYicKBIudKBBF77NbvC10LV7PderUqWZu9aoBv5duaWho8D8pAetE0j179phjJ06caObvvvuumV955ZVmbi3X9Prs3kmrXu710i1eH93jnQJr5UmuJ7HwmZ0oECx2okCw2IkCwWInCgSLnSgQLHaiQLDYiQJRUn12b123ZdasWWa+YcMGM7e29gWAZcuWxWaXX365Ofaqq64y82PHjpm5d2SztVW195iuXr3azDOZjJl72x5bNm/ebOZz5841846ODjO3tntO+yhrq5furePPVaJiF5EWAEcBnAFwWlXt7wwiSk0+ntmvUdXOPHwdIiog/s5OFIikxa4A3hWRT0Vk8UCfICKLRSQrIlnvyB0iKpykxT5PVX8I4AYA94rI1d/+BFVdqqoZVc2MGzcu4d0RUa4SFbuqtkZvOwCsAHBpPiZFRPmXc7GLSKWIVJ19H8DPANj9LSJKTZJX42sBrIh6gkMA/K+qvu0Nso6b9frNSY429vrwHq/fnITXR/dYa6O7u7vNsd569xdeeMHMb7nlFjOfN29ebDZ9+nRzbE9Pj5mPHz/ezHfu3GnmaWptbY3NvD0GcpVzsavqDgD2VQ9EVDLYeiMKBIudKBAsdqJAsNiJAsFiJwpE0Ze4lpWVxWbe9ruhWr9+vZlfeOGFsVl9fb051lv6e8UVV5j5/PnzzTxNkyZNynlsS0uLmTc2Npr5ypUrzfy6666Lzbx/E2tbc2tZL5/ZiQLBYicKBIudKBAsdqJAsNiJAsFiJwoEi50oECW1lXRFRYWZr127Njbzth32tsTytkT+7LPPYjPv2OKkZs+enfNY71jk9vZ2M/e22PaOurbGe9dVWFtkA/6SZ2t5r/fv3dvba+aeq68+Z9OmQfOWY1v/ptY21HxmJwoEi50oECx2okCw2IkCwWInCgSLnSgQLHaiQJRUn93r2Xq9dMuYMWPM3Ov5FrKX/sEHH5i5t2XykSNHYjPr+gDA7yc/+OCDZl5dXW3mVh/fOyHo8OHDiXLr7/b22/au5++8846ZP/vss2buXZ9gXUPgXftQW1sbm7HPTkQsdqJQsNiJAsFiJwoEi50oECx2okCw2IkCIapatDvLZDKazWaLdn/5ZM3bO87ZOzrY66N76/xfffXV2Ozmm282xxbajh07YrOGhgZzbNJzBDZu3BibzZw50xy7adMmM+/q6jLzOXPmmLn1/dTU1GSOta4ZyWQyyGazAzbb3Wd2EXleRDpEZEO/26pFZKWIbI3enu99HSJK12B+jP8rgOu/ddtDAFap6lQAq6KPiaiEucWuqqsBfPs61psALI/eXw4g3Z8ViciV6wt0taraFr2/D0DsxboislhEsiKS9faBI6LCSfxqvPa9whf7Kp+qLlXVjKpmvIUPRFQ4uRZ7u4jUAUD0tiN/UyKiQsi12F8HsCh6fxGA1/IzHSIqFLfPLiIvAvgxgBoA7QB+D+BVAP8EcCGAnQBuV1V7MTr8Prt1tjQAnHde/P9N3vphb/90bx9xy759+8x8woQJZn78+HEzf+mll8z8zjvvjM3ee+89c+y1115r5p5du3aZuXV2fFtbW2wGAHV1dTnNKR+875chQ+ytIE6dOmXmSa8hiGP12d3NK1T1jpjop4lmRURFxctliQLBYicKBIudKBAsdqJAsNiJAlFSW0l7WwOff3784rry8nJz7N69e83ca71t3749NmtsbDTHekaMGGHmCxcuNPP169fHZhdccEFOczrLW+qZpD3W2tpasK8N2Md0e1dzeq01T6Faa4B9FLXVvuYzO1EgWOxEgWCxEwWCxU4UCBY7USBY7ESBYLETBaKoffYzZ87g0KFDZp4rb3ls0n7zlClTch7r9ZPr6+vN3DreFwDKyspis1deecUc6/X4Z8yYYealzHpcPEmXuG7evNnMW1paYrP58+ebY60asubNZ3aiQLDYiQLBYicKBIudKBAsdqJAsNiJAsFiJwpEUfvsZWVl5nGzXq/c6sN7PdW1a9ea+dy5c808Ca+P7tm6dauZd3Z2xmZLlixJdN+eLVu2mPm0adNiM++oa4+3/0F1dXXOX9vbHnzixIlmPn36dDNPMjdrS3WRAXeR7huX8z0S0fcKi50oECx2okCw2IkCwWInCgSLnSgQLHaiQJTUvvFW/9Cze/duMy9kH907cnnHjh1mPmvWLDN/7LHHzPySSy6Jzbxe9G233WbmHquP7vHm9sADD5j5k08+aebWceRWPxrw++ieI0eOmLm3b73F2pvB2q/erS4ReV5EOkRkQ7/bHhWRVhFpjv4s+K4TJqLiGsxT6V8BXD/A7X9S1aboz1v5nRYR5Ztb7Kq6GsDBIsyFiAooyQt094nIuujH/NhD2ERksYhkRSRrnb1FRIWVa7H/BcAUAE0A2gD8Ie4TVXWpqmZUNZPkRQkiSianYlfVdlU9o6pfAXgWwKX5nRYR5VtOxS4i/c/S/TmADXGfS0SlQaxeJACIyIsAfgygBkA7gN9HHzcBUAAtAH6tqm3enWUyGc1ms7F5V1eXOd46nz0pb896a718c3OzOXbSpElmXlVVZebeWd8bN26Mzbx93zdssP+f9q4BKKQ33njDzL11/vfff39s5j2mBw4cMPOxY8eaucfa3z3J2fCZTAbZbHbAiwjcr6qqdwxw87KcZ0NEqeDlskSBYLETBYLFThQIFjtRIFjsRIEoqSWuSezZs8fMvSWL3jbWVutt8uTJ5thRo0aZueeuu+4y8yQtyfLy8pzHAsCaNWty/vrPPfecOfbuu+8284ULF5p5EtaxyEDy1pvVXvOWa+d6/Dif2YkCwWInCgSLnSgQLHaiQLDYiQLBYicKBIudKBAl1WdP0i+eMGFCovv2ljwWauxgPPLII2Y+fvz42OzLL780x1588cVm3tvba+beNQTW17/11lvNsV988YWZe98v1s5I3tJebws17+/tbVX95ptvxmaLFi0yx1rXAFhLtfnMThQIFjtRIFjsRIFgsRMFgsVOFAgWO1EgWOxEgShqn11VcfLkydjcO7LZ6md7W/9aW/cCfj+5vr4+Ntu3b585trGx0cy9nqy3vtnaqtpahw8A7e3tZj5s2DAz37Vrl5mvWLEiNnvqqafMsU8//bSZL1myxMyt9fAnTpwwx86ePdvMR48ebeZ79+41c6+XbhkzZkxsZv1785mdKBAsdqJAsNiJAsFiJwoEi50oECx2okCw2IkCUfT17FYv/ciRI+ZYa69ub32x97WtPjoAbNu2LTbz1l1XVlaaubUeHQDmzZtn5paKigozt9Z8A0BnZ6eZX3bZZWb+k5/8JDazHlMAaGhoMHNvPbu1Jt3qVQPAwYMHzfzjjz828wULFph5GtxndhFpEJH3RWSjiHwuIr+Jbq8WkZUisjV6W7jD04koscH8GH8awO9UdSaAywHcKyIzATwEYJWqTgWwKvqYiEqUW+yq2qaqa6L3jwLYBKAewE0AlkefthzAzYWaJBEl951eoBORRgA/APBvALWq2hZF+wDUxoxZLCJZEcl6v/8RUeEMuthFZCSAVwD8VlW/8WqXqioAHWicqi5V1YyqZmpqahJNlohyN6hiF5Gh6Cv0F1T1X9HN7SJSF+V1ADoKM0Uiyge39SZ96y+XAdikqn/sF70OYBGAJ6K3rw3mDq2jkb02kcVbyllbO+BvGV/r6ekxc2tLZG8Z6NGjR83cayG1traaubWE1tpaGPCXBnd1dZm599Pa5s2bY7NnnnnGHOstv3388cdzvm9vC+0PP/zQzL3WmnXfADB9+nQzt1jLZ0+dOhWbDabP/iMAvwSwXkSao9seRl+R/1NEfgVgJ4DbBztZIio+t9hV9QMAcbsr/DS/0yGiQuHlskSBYLETBYLFThQIFjtRIFjsRIEo+hJXa9vk48ePm2NHjBgRm3lLEr0jnb0e/5YtW2Kziy66yBz70UcfmfnEiRPN3NuK2tqS+Z577jHHerw+vLe81/o3ffnll82x3rUR3d3dZm5d37BmzRpzrLf0t7m52cybmprM/NixY7GZtyTa23I9dlxOo4joe4fFThQIFjtRIFjsRIFgsRMFgsVOFAgWO1EgpG+TmeLIZDKazWYL8rW9I3i93qS1zh4AysvLYzNrDTFg91QBoKqqysy9fyPrGgDr2gTA7+GXsnXr1pn5nDlzYrP333/fHHvNNdfkNKd88PYgsPZuyGQyyGazA17Mwmd2okCw2IkCwWInCgSLnSgQLHaiQLDYiQLBYicKRNHXsxeK1QcfjN7eXjO31kYn7ZN7e957Zs6cmWj895XVR/ck7aOfPn3azL2zAqzrOrzrNry9GeLwmZ0oECx2okCw2IkCwWInCgSLnSgQLHaiQLDYiQIxmPPZGwD8DUAtAAWwVFX/LCKPArgLwP7oUx9W1bcKNdGkvD66d8a6l1u889dL2bZt28zc6/mOHDkyn9MpGUOG2KVTiv/mg7mo5jSA36nqGhGpAvCpiKyMsj+p6n8XbnpElC+DOZ+9DUBb9P5REdkEoL7QEyOi/PpOv7OLSCOAHwD4d3TTfSKyTkSeF5EBf24RkcUikhWR7P79+wf6FCIqgkEXu4iMBPAKgN+q6hEAfwEwBUAT+p75/zDQOFVdqqoZVc1452cRUeEMqthFZCj6Cv0FVf0XAKhqu6qeUdWvADwL4NLCTZOIknKLXfqOXV0GYJOq/rHf7XX9Pu3nADbkf3pElC+DeTX+RwB+CWC9iJw9p/ZhAHeISBP62nEtAH49mDu0tslNutTTkqR1BhR23t5ySK/NYy2JHDVqVKL7njx5spl7W3RbxyonbcsdOnTIzMeMGRObWdtvA8C0adNymtNZ3nHSabQkB/Nq/AcABtqHumR76kR0Ll5BRxQIFjtRIFjsRIFgsRMFgsVOFAgWO1Egir6VtNWT9rbQHTp0aGzmHbnc09Nj5pWVlWZu8e7b60VXVFSYuddn98ZbvG2wT548aebDhw8380IeCW710T1J++ger4+e5PoD69oI63uRz+xEgWCxEwWCxU4UCBY7USBY7ESBYLETBYLFThQIKWQf9Jw7E9kPYGe/m2oAdBZtAt9Nqc6tVOcFcG65yufcJqnqgPu/FbXYz7lzkayqZlKbgKFU51aq8wI4t1wVa278MZ4oECx2okCkXexLU75/S6nOrVTnBXBuuSrK3FL9nZ2IiiftZ3YiKhIWO1EgUil2EbleRDaLyDYReSiNOcQRkRYRWS8izSKSTXkuz4tIh4hs6HdbtYisFJGt0dtUzgaOmdujItIaPXbNIrIgpbk1iMj7IrJRRD4Xkd9Et6f62BnzKsrjVvTf2UWkDMAWANcB2APgEwB3qOrGok4khoi0AMioauoXYIjI1QC6AfxNVWdFt/0XgIOq+kT0H+X5qvqfJTK3RwF0p32Md3RaUV3/Y8YB3AzgP5DiY2fM63YU4XFL45n9UgDbVHWHqvYC+AeAm1KYR8lT1dUADn7r5psALI/eX46+b5aii5lbSVDVNlVdE71/FMDZY8ZTfeyMeRVFGsVeD2B3v4/3oLTOe1cA74rIpyKyOO3JDKBWVdui9/cBqE1zMgNwj/Eupm8dM14yj10ux58nxRfozjVPVX8I4AYA90Y/rpYk7fsdrJR6p4M6xrtYBjhm/GtpPna5Hn+eVBrF3gqgod/HE6PbSoKqtkZvOwCsQOkdRd1+9gTd6G1HyvP5Wikd4z3QMeMogccuzePP0yj2TwBMFZGLRGQYgF8AeD2FeZxDRCqjF04gIpUAfobSO4r6dQCLovcXAXgtxbl8Q6kc4x13zDhSfuxSP/5cVYv+B8AC9L0ivx3AI2nMIWZekwGsjf58nvbcALyIvh/rTqHvtY1fARgLYBWArQDeA1BdQnP7O4D1ANahr7DqUprbPPT9iL4OQHP0Z0Haj50xr6I8brxcligQfIGOKBAsdqJAsNiJAsFiJwoEi50oECx2okCw2IkC8X/iys4Vklv1EQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gyp9nSDgx8gq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "5914c0f2-7507-45bf-8daf-0053e76f7a50"
      },
      "source": [
        "plt.imshow(introspection_testing_images[1,:,:,0], cmap = \"binary\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fc2b2fef630>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DUlF-AKHoju1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_figures(introspection_testing_images, masked_grads, num_images=25):\n",
        "  x_subplots = tf.math.sqrt(tf.constant(num_images, dtype = tf.float32))\n",
        "  plt.figure(figsize=(10,10))\n",
        "  j = -1\n",
        "  for i in range(0, num_images):\n",
        "      plt.subplot(6, 6,i+1)\n",
        "      plt.xticks([])\n",
        "      plt.yticks([])\n",
        "      plt.grid(False)\n",
        "      j = j + 1\n",
        "      if(i % 2 == 0):  \n",
        "        plt.imshow(introspection_testing_images[i, :, :, 0], cmap=plt.cm.binary)\n",
        "        #print(i)\n",
        "      elif(j%2 == 1):      \n",
        "        plt.imshow(tf.nn.relu(masked_grads[0][j-1, :, :, 0]), cmap=\"Greys_r\")\n",
        "        #print(j-1)\n",
        "      \n",
        "      #plt.xlabel(class_names[test_images[i]])\n",
        "  plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_EsPKQR2sPdU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "d7696223-3e01-4d0b-ff67-43c161291309"
      },
      "source": [
        "plt.imshow(masked_grads[0][8, :, :, 0], cmap=plt.cm.binary)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fc2b2f527f0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAKZElEQVR4nO3dT4ic933H8fendnJxcpCrRQjHVGkwBVOoEhZRiAkpaYLti5xLiA5BBYNysCGBHGrSQ300oUnooQSUWkQtqUMhMdbBtHFFwARK8NqotmzTyjUKkZClET7EOSV2vj3s47CRd7XrmWf+0O/7BcvMPPPsPl8GvzUzzwz+paqQ9P/fHyx7AEmLYexSE8YuNWHsUhPGLjVx6yIPtn///jp06NAiDym1cvHiRa5fv57t7psp9iT3An8P3AL8Y1U9drP9Dx06xMbGxiyHlHQT6+vrO9439cv4JLcA/wDcB9wNHEty97R/T9J8zfKe/QjwWlW9XlW/Bn4AHB1nLEljmyX2O4BfbLl9adj2e5KcSLKRZGMymcxwOEmzmPvZ+Ko6WVXrVbW+trY278NJ2sEssV8G7txy+yPDNkkraJbYnwPuSvLRJB8EvgicGWcsSWOb+qO3qno7ycPAv7P50dupqnp5tMkkjWqmz9mr6mng6ZFmkTRHfl1WasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJmZasjnJReAt4B3g7apaH2MoSeObKfbBX1TV9RH+jqQ58mW81MSssRfw4yTPJzmx3Q5JTiTZSLIxmUxmPJykac0a+z1V9QngPuChJJ+6cYeqOllV61W1vra2NuPhJE1rptir6vJweQ14EjgyxlCSxjd17EluS/Lhd68DnwPOjzWYpHHNcjb+APBkknf/zr9U1b+NMpWk0U0de1W9DvzZiLNImiM/epOaMHapCWOXmjB2qQljl5owdqkJY5eaMHapCWOXmjB2qQljl5owdqkJY5eaMHapCWOXmjB2qQljl5owdqkJY5eaMHapCWOXmjB2qQljl5owdqkJY5eaMHapCWOXmjB2qQljl5owdqmJXWNPcirJtSTnt2y7PckzSS4Ml/vmO6akWe3lmf17wL03bHsEOFtVdwFnh9uSVtiusVfVs8CbN2w+Cpwerp8GHhh5Lkkjm/Y9+4GqujJcfwM4sNOOSU4k2UiyMZlMpjycpFnNfIKuqgqom9x/sqrWq2p9bW1t1sNJmtK0sV9NchBguLw23kiS5mHa2M8Ax4frx4GnxhlH0rzs5aO3J4D/BP4kyaUkDwKPAZ9NcgH4y+G2pBV26247VNWxHe76zMizSJojv0EnNWHsUhPGLjVh7FITxi41YexSE8YuNWHsUhPGLjVh7FITxi41YexSE8YuNWHsUhPGLjVh7FITxi41YexSE8YuNWHsUhPGLjVh7FITxi41YexSE8YuNWHsUhPGLjVh7FITxi41YexSE3tZn/1UkmtJzm/Z9miSy0nODT/3z3dMSbPayzP794B7t9n+7ao6PPw8Pe5Yksa2a+xV9Szw5gJmkTRHs7xnfzjJi8PL/H077ZTkRJKNJBuTyWSGw0maxbSxfwf4GHAYuAJ8c6cdq+pkVa1X1fra2tqUh5M0q6lir6qrVfVOVf0W+C5wZNyxJI1tqtiTHNxy8/PA+Z32lbQabt1thyRPAJ8G9ie5BPwt8Okkh4ECLgJfnuOMkkawa+xVdWybzY/PYRZJc+Q36KQmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWpi19iT3JnkJ0leSfJykq8M229P8kySC8PlvvmPK2lae3lmfxv4WlXdDfw58FCSu4FHgLNVdRdwdrgtaUXtGntVXamqF4brbwGvAncAR4HTw26ngQfmNaSk2b2v9+xJDgEfB34GHKiqK8NdbwAHdvidE0k2kmxMJpMZRpU0iz3HnuRDwA+Br1bVL7feV1UF1Ha/V1Unq2q9qtbX1tZmGlbS9PYUe5IPsBn696vqR8Pmq0kODvcfBK7NZ0RJY9jL2fgAjwOvVtW3ttx1Bjg+XD8OPDX+eJLGcuse9vkk8CXgpSTnhm1fBx4D/jXJg8DPgS/MZ0RJY9g19qr6KZAd7v7MuONImhe/QSc1YexSE8YuNWHsUhPGLjVh7FITxi41YexSE8YuNWHsUhPGLjVh7FITxi41YexSE8YuNWHsUhPGLjVh7FITxi41YexSE8YuNWHsUhPGLjVh7FITxi41YexSE8YuNWHsUhPGLjWxl/XZ70zykySvJHk5yVeG7Y8muZzk3PBz//zHlTStvazP/jbwtap6IcmHgeeTPDPc9+2q+rv5jSdpLHtZn/0KcGW4/laSV4E75j2YpHG9r/fsSQ4BHwd+Nmx6OMmLSU4l2bfD75xIspFkYzKZzDSspOntOfYkHwJ+CHy1qn4JfAf4GHCYzWf+b273e1V1sqrWq2p9bW1thJElTWNPsSf5AJuhf7+qfgRQVVer6p2q+i3wXeDI/MaUNKu9nI0P8DjwalV9a8v2g1t2+zxwfvzxJI1lL2fjPwl8CXgpyblh29eBY0kOAwVcBL48lwkljWIvZ+N/CmSbu54efxxJ8+I36KQmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qwtilJoxdasLYpSaMXWrC2KUmjF1qIlW1uIMlE+DnWzbtB64vbID3Z1VnW9W5wNmmNeZsf1RV2/7/3xYa+3sOnmxU1frSBriJVZ1tVecCZ5vWombzZbzUhLFLTSw79pNLPv7NrOpsqzoXONu0FjLbUt+zS1qcZT+zS1oQY5eaWErsSe5N8t9JXkvyyDJm2EmSi0leGpah3ljyLKeSXEtyfsu225M8k+TCcLntGntLmm0llvG+yTLjS33slr38+cLfsye5Bfgf4LPAJeA54FhVvbLQQXaQ5CKwXlVL/wJGkk8BvwL+qar+dNj2DeDNqnps+IdyX1X99YrM9ijwq2Uv4z2sVnRw6zLjwAPAX7HEx+4mc32BBTxuy3hmPwK8VlWvV9WvgR8AR5cwx8qrqmeBN2/YfBQ4PVw/zeZ/LAu3w2wroaquVNULw/W3gHeXGV/qY3eTuRZiGbHfAfxiy+1LrNZ67wX8OMnzSU4se5htHKiqK8P1N4ADyxxmG7su471INywzvjKP3TTLn8/KE3TvdU9VfQK4D3hoeLm6kmrzPdgqfXa6p2W8F2WbZcZ/Z5mP3bTLn89qGbFfBu7ccvsjw7aVUFWXh8trwJOs3lLUV99dQXe4vLbkeX5nlZbx3m6ZcVbgsVvm8ufLiP054K4kH03yQeCLwJklzPEeSW4bTpyQ5Dbgc6zeUtRngOPD9ePAU0uc5fesyjLeOy0zzpIfu6Uvf15VC/8B7mfzjPz/An+zjBl2mOuPgf8afl5e9mzAE2y+rPsNm+c2HgT+EDgLXAD+A7h9hWb7Z+Al4EU2wzq4pNnuYfMl+ovAueHn/mU/djeZayGPm1+XlZrwBJ3UhLFLTRi71ISxS00Yu9SEsUtNGLvUxP8By+ZJMMx9nEMAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EIXLXp52uOit",
        "colab_type": "text"
      },
      "source": [
        "This is the side by side comparision of images with mask of number one applied\n",
        "\n",
        "- We can see that numbers **6, 8, 0** which doesn't have stick shape has no bearing or contribution of softmax(even a little softmax of one is not present in these numbers). S**o these numbers are insensitive to the number one.**\n",
        "\n",
        "- For some numbers like **3, 9 and 4** we can see a slight presence of one when the numbers are slanted. **Less sensitivity**\n",
        "\n",
        "- For number **7** the presence of softmax ones contribution can be clearly seen especially at last image. **Some sensitivity.**\n",
        "\n",
        "- **For contribution of one to one itself, we can see some strong sensitivity.** But thing to note here is for the second one(the one in 3rd row which is really straight and not slant) **it is surprising to see that one is getting recognized due to sensitivity of irrelvenat pixels rather the stick like figure, for one that are slant the activations are from stick.** **Same pattern is present in 7 too if we can see at the last 7 where slnated 7 is getting recognized.**\n",
        "\n",
        "- **This clearly demonstrates that CNN based detection models are non-human like, even though we think them to be. Even after 50 - 60k examples CNN was not able to learn the context better than humans where humans have lower resolution than machines. This is a clear demonstration that we need Better models than CNNs in the future**."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BilQ4oROqMEP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 577
        },
        "outputId": "22cbf0c7-efe1-488c-9d9a-8c3488de9895"
      },
      "source": [
        "plot_figures(introspection_testing_images, masked_grads, num_images=36)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 36 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4tBmaWUmxfBl",
        "colab_type": "text"
      },
      "source": [
        "**Sensitivity analysis for number 8**\n",
        "\n",
        "- Surprisingly number 1 and 7 which are least related to number 8 are getting activated. Seems like slanted stick like pattern is getting activated everytime.\n",
        "\n",
        "- Numbers 9 are getting activated which are no surprise as upper part of 9 looks like a half of 8. But 9 was activated mainly due to its lower part and garbage pixels. which raises doubts on ability of CNN to understand semantic meaning of images.\n",
        "\n",
        "- 3 should have been activated as it looks like a half of 8, but it dod not happen."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LvBVCSTEw9hO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "tape, masked_loss = record_tape(introspection_testing_images, introspection_testing_labels, mask_list=[eight_mask])\n",
        "masked_grads = calculate_masked_grads(introspection_testing_images, introspection_testing_labels, tape, masked_loss)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OrEzb-ohxRsx",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 577
        },
        "outputId": "5d49accd-0efa-4401-9d13-2af91de88624"
      },
      "source": [
        "plot_figures(introspection_testing_images, masked_grads, num_images=36)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 36 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fmSIB_Dufl9j",
        "colab_type": "text"
      },
      "source": [
        "# Activation Maximization\n",
        "\n",
        "**Summary**\n",
        "\n",
        "- Due to black and white natureof mnsit image, the bright colored high frequency colors are not present. Hence the difference visible is very less.\n",
        "\n",
        "- However when we take a difference between modified image and actual image, we see a change in order of e-4. This along with increase in correct classifications of 1 t from 550 to 606 confirms the activation maximization"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "E1-e6iGlDLTl",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        },
        "outputId": "6507b833-d0d0-4695-a7d2-90b583a9adf9"
      },
      "source": [
        "!unzip \"/content/vgg_cnn_model_introspection.zip\" "
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Archive:  /content/vgg_cnn_model_introspection.zip\n",
            "   creating: content/vgg_cnn_model_introspection/\n",
            "  inflating: content/vgg_cnn_model_introspection/saved_model.pb  \n",
            "   creating: content/vgg_cnn_model_introspection/variables/\n",
            "  inflating: content/vgg_cnn_model_introspection/variables/variables.data-00001-of-00002  \n",
            "  inflating: content/vgg_cnn_model_introspection/variables/variables.index  \n",
            "  inflating: content/vgg_cnn_model_introspection/variables/variables.data-00000-of-00002  \n",
            "   creating: content/vgg_cnn_model_introspection/assets/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EE5o6hWTDMLy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vgg_cnn_model = keras.models.load_model(filepath=\"/content/content/vgg_cnn_model_introspection\")"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MxSSC-Qdfq9L",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "848df356-c529-4af7-8365-bad34ca29e82"
      },
      "source": [
        "adverserial_training_images_batch = tf.convert_to_tensor(train_images[0:1024, :, :], dtype=tf.float32)\n",
        "adverserial_test_images_batch = tf.convert_to_tensor(test_images[0:1024, :, :], dtype=tf.float32)\n",
        "adverserial_training_labels_batch = tf.convert_to_tensor(train_labels[0:1024], dtype=tf.float32)\n",
        "adverserial_test_labels_batch = tf.convert_to_tensor(test_labels[0:1024], dtype=tf.float32)\n",
        "print(adverserial_training_images_batch.shape)\n",
        "print(adverserial_test_images_batch.shape)\n",
        "print(adverserial_training_labels_batch.shape)\n",
        "print(adverserial_test_labels_batch.shape)"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(1024, 28, 28)\n",
            "(1024, 28, 28)\n",
            "(1024,)\n",
            "(1024,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yRgKH1hyfrAv",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 170
        },
        "outputId": "2dfc388c-b30c-44a2-fbe4-8800371cedaf"
      },
      "source": [
        "test_onehot = tf.one_hot(indices = tf.cast(adverserial_test_labels_batch, dtype=tf.dtypes.int64), depth=10)\n",
        "print(test_onehot)\n",
        "test_onehot.shape"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tf.Tensor(\n",
            "[[0. 0. 0. ... 1. 0. 0.]\n",
            " [0. 0. 1. ... 0. 0. 0.]\n",
            " [0. 1. 0. ... 0. 0. 0.]\n",
            " ...\n",
            " [0. 0. 0. ... 1. 0. 0.]\n",
            " [0. 0. 0. ... 0. 0. 0.]\n",
            " [0. 0. 0. ... 0. 0. 0.]], shape=(1024, 10), dtype=float32)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "TensorShape([1024, 10])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r4MAFMOfCrei",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "zero_mask = tf.tile(tf.Variable(initial_value=tf.constant([1,0,0,0,0,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "one_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,1,0,0,0,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "two_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,1,0,0,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "three_mask = tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,1,0,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "four_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,1,0,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "five_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,1,0,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "six_mask = tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,0,1,0,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "seven_mask = tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,0,0,1,0,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "eight_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,0,0,0,1,0], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])\n",
        "nine_mask =  tf.tile(tf.Variable(initial_value=tf.constant([0,0,0,0,0,0,0,0,0,1], dtype=\"float32\", shape=(1,10))) , multiples=[1024,1])"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7esQ-JSjAWQk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def create_adversarial_pattern(adverserial_training_images_batch, adverserial_training_labels_batch, perturbation_type=\"multiply\", \n",
        "                               attack_type = \"untargeted\", targeted_attack_images = None, loss_fnc = \"CCE\", lmbda = 0.05, multiply_factor=1000000, mask = one_mask):\n",
        "  with tf.GradientTape() as tape:\n",
        "    tape.watch(adverserial_training_images_batch)\n",
        "    prediction = vgg_cnn_model(adverserial_training_images_batch)\n",
        "    masked_prediction = tf.multiply(mask, prediction)\n",
        "    #print(masked_prediction[0:5,:])\n",
        "    #Two loss functions for targeted and untargeted, used masked loss\n",
        "    if(attack_type == \"untargeted\" and loss_fnc == \"CCE\"):\n",
        "      loss = tf.keras.losses.sparse_categorical_crossentropy(adverserial_training_labels_batch, masked_prediction)\n",
        "    elif(attack_type == \"untargeted\" and loss_fnc == \"MSE\"):\n",
        "      one_hot_test_labels = tf.one_hot(indices = tf.cast(adverserial_training_labels_batch, dtype=tf.dtypes.int64), depth=10)\n",
        "      loss = tf.keras.losses.mean_squared_error(one_hot_test_labels, prediction)\n",
        "    elif(attack_type == \"targeted\" and loss_fnc == \"MSE\"):\n",
        "      one_hot_test_labels = tf.one_hot(indices = tf.cast(adverserial_training_labels_batch, dtype=tf.dtypes.int64), depth=10)\n",
        "      \n",
        "      y_loss = lmbda * tf.keras.losses.mean_squared_error(adverserial_training_images_batch, targeted_attack_images)\n",
        "      \n",
        "      targeting_image_loss = tf.reduce_sum(y_loss,axis = 1)\n",
        "      \n",
        "      loss = tf.keras.losses.mean_squared_error(one_hot_test_labels, prediction) + targeting_image_loss\n",
        "    else:\n",
        "      print(\"Only two types of attacks are possible untargeted and targeted, you have given {}. Aborting\".format(attack_type))\n",
        "      return [None, None]\n",
        "\n",
        "  # Get the gradients of the loss w.r.t to the input image.\n",
        "  #print(loss[0:5])\n",
        "  gradient = tape.gradient(loss,adverserial_training_images_batch)\n",
        "  # Get the sign of the gradients to create the perturbation\n",
        "  #print(gradient.shape)\n",
        "  if(perturbation_type == \"multiply\"):\n",
        "    perturbations = multiply_factor * gradient\n",
        "  elif(perturbation_type == \"unit_vector\"):\n",
        "    z = gradient\n",
        "    norm = tf.sqrt(tf.reduce_sum(tf.square(z), axis=[-1,-2], keepdims=True))\n",
        "    perturbations = z/norm\n",
        "  elif(perturbation_type == \"gradient_sign\"):\n",
        "    perturbations = tf.sign(gradient)\n",
        "  else:\n",
        "    perturbations = gradient\n",
        "  return [gradient, perturbations];"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RinVq3wGAWVy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 238
        },
        "outputId": "32bd2ab4-0792-4df4-b704-9740eca0874f"
      },
      "source": [
        "gradient, perturbations = create_adversarial_pattern(adverserial_training_images_batch, adverserial_training_labels_batch, mask=one_mask)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tf.Tensor(\n",
            "[[0.0000000e+00 5.4422884e-07 0.0000000e+00 0.0000000e+00 0.0000000e+00\n",
            "  0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00]\n",
            " [0.0000000e+00 2.0251712e-11 0.0000000e+00 0.0000000e+00 0.0000000e+00\n",
            "  0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00]\n",
            " [0.0000000e+00 6.1249075e-06 0.0000000e+00 0.0000000e+00 0.0000000e+00\n",
            "  0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00]\n",
            " [0.0000000e+00 9.9999344e-01 0.0000000e+00 0.0000000e+00 0.0000000e+00\n",
            "  0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00]\n",
            " [0.0000000e+00 1.4561766e-03 0.0000000e+00 0.0000000e+00 0.0000000e+00\n",
            "  0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00 0.0000000e+00]], shape=(5, 10), dtype=float32)\n",
            "tf.Tensor([2.6701605e+00 2.3025851e+00 4.2520466e+00 1.0728830e-06 9.5867729e+00], shape=(5,), dtype=float32)\n",
            "(1024, 28, 28)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oA5r49s-AWkK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "c6d2c103-842c-4919-8361-7a1fa58fcb9c"
      },
      "source": [
        "perturbations.shape"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "TensorShape([1024, 28, 28])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "51eBYd1WELw0",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "b7d05ff0-5b92-4d12-ae1c-af9e3a768ca4"
      },
      "source": [
        "plt.imshow(adverserial_training_images_batch[3,:,:], cmap=plt.cm.binary)"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7feb7ab1bda0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w9zr2FFnAfiF",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "plt.imshow(perturbations[0,:,:], cmap=plt.cm.binary)\n",
        "print(perturbations[3,:,:])\n",
        "print(gradient[3:, :, :])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yxsmp2M8Afnw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def activation_maximization(adverserial_training_images_batch, adverserial_training_labels_batch,  perturbation_type=\"multiply\", index_of_num=3,\n",
        "                      attack_type = \"untargeted\", targeted_attack_images = None, loss_fnc = \"CCE\"):\n",
        "  \n",
        "  plt.figure(figsize=(10,10))\n",
        "  for i in range(25):\n",
        "    plt.subplot(5,5,i+1)\n",
        "    plt.xticks([])\n",
        "    plt.yticks([])\n",
        "    plt.grid(False)\n",
        "    \n",
        "    for j in range(20):\n",
        "      #Get the weighted gradient/maximum change of the individual image pixel wrt output prediction. weighted by 0.071\n",
        "      if(i == 0):\n",
        "        gradient, perturbations = create_adversarial_pattern(adverserial_training_images_batch, adverserial_training_labels_batch, perturbation_type, attack_type, loss_fnc=loss_fnc)\n",
        "      elif(i > 0):\n",
        "        gradient, perturbations = create_adversarial_pattern(adverserial_training_images_batch_modified, modified_predictions_2, perturbation_type, attack_type, loss_fnc=loss_fnc)\n",
        "\n",
        "      #Add the gradient instead of subtracting it as we are trying to maximize the loss and moving in opposite direction of convex optimization hill.\n",
        "      if(i == 0):\n",
        "        adverserial_training_images_batch_modified = tf.subtract(adverserial_training_images_batch, 1 * perturbations)\n",
        "      elif(i > 0):\n",
        "        adverserial_training_images_batch_modified = tf.subtract(adverserial_training_images_batch_modified, 1 * perturbations)\n",
        "\n",
        "      #Get the new predictions for each change\n",
        "      modified_predictions = vgg_cnn_model(adverserial_training_images_batch_modified)\n",
        "      modified_predictions_2 =  tf.cast(tf.argmax(modified_predictions, axis=1),  dtype=tf.dtypes.float32)\n",
        "\n",
        "      #Get the number of correct classifications to monitor misclassifications and see gradual changes\n",
        "      x = tf.equal(adverserial_training_labels_batch, modified_predictions_2)\n",
        "      print(\"Number of correct classifications, prediction of first image and counter are: {}, {} and {}\".format(tf.reduce_sum(tf.cast(x, tf.int32)), modified_predictions_2[index_of_num], i))\n",
        "      #print(\"Prediction of first image is : {}\".format(modified_predictions_2[0]))\n",
        "    plt.imshow(adverserial_training_images_batch_modified[index_of_num], cmap=plt.cm.binary)\n",
        "  \n",
        "  return adverserial_training_images_batch_modified;"
      ],
      "execution_count": 39,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BsJVB08LAfk7",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "686654f5-f089-4941-d6ef-daf86ba9912f"
      },
      "source": [
        "adverserial_training_images_batch_modified = activation_maximization(adverserial_training_images_batch, adverserial_training_labels_batch,  perturbation_type=\"multiply\")"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 0\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 1\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 2\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 3\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 4\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 5\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 6\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 7\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 8\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 9\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 10\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 11\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 12\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 13\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 14\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 15\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 16\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 17\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 18\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 19\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 20\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 21\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 22\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 23\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n",
            "Number of correct classifications, prediction of first image and counter are: 606, 1.0 and 24\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAj0AAAIxCAYAAACmUwl6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAZmElEQVR4nO3dMWtTbxsH4Jw3gouggxJxaeliP4AgCKJUcHTwA0hdleIkdBAnBx1c3AX9Bg4dLDooOrYg+AXapSD/QIubQznv9rSnNJLEnKa57+ua7kNOwgM/Aj+e57Sp6rruAABE979pLwAA4CQoPQBACkoPAJCC0gMApKD0AAApKD0AQApnRrn54sWL9fz8fEtL4ThbW1udfr9fTfpzZTkdm5ub/bquL036c+V58nw3Y2njuynL6fhbliOVnvn5+c7GxsZkVsVQrl271srnynI6qqrabuNz5XnyfDdjaeO7Kcvp+FuWjrcAgBSUHgAgBaUHAEhB6QEAUlB6AIAUlB4AIAWlBwBIQekBAFJQegCAFJQeACAFpQcASEHpAQBSUHoAgBSUHgAghTPTXsBp9uLFi8b18+fPy1zXdZm/fPnSuO/WrVutrovRyTIWecYhyzhmIUs7PQBACkoPAJCC460j3r17V+aXL182Xut2u2Xe398vc1VVra+L0ckyFnnGIcs4Zi1LOz0AQApKDwCQguOtI7a3t8v858+fKa6EfyXLWOQZhyzjmLUs7fQAACkoPQBACkoPAJCCZ3o6nc7nz5/L/ObNm4H3LS4ulnltba3MvV6vnYUxMlnGIs84ZBnHLGdppwcASEHpAQBSSHm89f3798b18vJymX///j3wfU+fPi3z3NzcxNfF6GQZizzjkGUckbK00wMApKD0AAApKD0AQAopn+l5//5943pnZ+fY+27fvt24fvDgQVtLYkyyjEWeccgyjkhZ2ukBAFJQegCAFNIcb/X7/TK/ffu28Vq32y3zhQsXyvzs2bP2F8bIZBmLPOOQZRxRs7TTAwCkoPQAACmEPt7a2toq8/3794d6z8rKSpmXlpYmvSTGJMtY5BmHLOPIkKWdHgAgBaUHAEhB6QEAUgj9TM/Hjx/L/PPnz4H33blzp8xPnjxpdU2MR5axyDMOWcaRIUs7PQBACkoPAJBCqOOtDx8+NK5XV1ePve/mzZuN68M/pnb+/PnJL4yRyTIWecYhyzgyZmmnBwBIQekBAFKY+eOtcf6D5MLCQuO61+tNckmMSZaxyDMOWcaRPUs7PQBACkoPAJCC0gMApDDzz/S8evWqzN1ud6j3DPqzPKZLlrHIMw5ZxpE9Szs9AEAKSg8AkMJMHm/9+PGjzOvr60O95969e2W+evXqxNfEeGQZizzjkGUcsjxgpwcASEHpAQBSmMnjrbt375Z5d3d34H3Xr18v8+EfSOP0kGUs8oxDlnHI8oCdHgAgBaUHAEhB6QEAUpjJZ3r6/X6Z//YfJR8/flzmc+fOtbomxiPLWOQZhyzjkOUBOz0AQApKDwCQwkwcbz18+LBxXdd1mff39we+78aNG62tifHIMhZ5xiHLOGQ5mJ0eACAFpQcASOHUHm8d/oG0T58+NV6rqqrMZ8+eLfOjR48a9/V6vZZWxyhkGYs845BlHLIcjp0eACAFpQcASEHpAQBSOLXP9Ozt7ZX5169fA++7cuVKmV+/ft3qmhiPLGORZxyyjEOWw7HTAwCkoPQAACkoPQBACkoPAJCC0gMApKD0AAApnNo/WV9cXCzz0V9+/fbt20kvh38gy1jkGYcs45DlcOz0AAApKD0AQAqn9njr8uXLZf769esUV8K/kmUs8oxDlnHIcjh2egCAFJQeACAFpQcASEHpAQBSUHoAgBSUHgAgBaUHAEhB6QEAUlB6AIAUqrquh7+5qv7rdDrb7S2HY8zVdX1p0h8qy6mRZxyyjGXiecpyagZmOVLpAQCYVY63AIAUlB4AIAWlBwBIQekBAFJQegCAFJQeACAFpQcASEHpAQBSUHoAgBSUHgAgBaUHAEhB6QEAUlB6AIAUlB4AIAWlBwBIQekBAFJQegCAFM6McvPFixfr+fn5lpbCcba2tjr9fr+a9OfKcjo2Nzf7dV1fmvTnyvPk+W7G0sZ3U5bT8bcsRyo98/PznY2NjcmsiqFcu3atlc+V5XRUVbXdxufK8+T5bsbSxndTltPxtywdbwEAKSg9AEAKSg8AkILSAwCkoPQAACkoPQBACkoPAJCC0gMApKD0AAApKD0AQApKDwCQgtIDAKSg9AAAKSg9AEAKSg8AkMKZaS/gNHvx4kXj+vnz52Wu67rMX758adx369atVtfF6GQZizzjkGUcs5ClnR4AIAWlBwBIwfHWEe/evSvzy5cvG691u90y7+/vl7mqqtbXxehkGYs845BlHLOWpZ0eACAFpQcASMHx1hHb29tl/vPnzxRXwr+SZSzyjEOWccxalnZ6AIAUlB4AIAWlBwBIwTM9nU7n8+fPZX7z5s3A+xYXF8u8trZW5l6v187CGJksY5FnHLKMY5aztNMDAKSg9AAAKaQ83vr+/Xvjenl5ucy/f/8e+L6nT5+WeW5ubuLrYnSyjEWeccgyjkhZ2ukBAFJQegCAFJQeACCFlM/0vH//vnG9s7Nz7H23b99uXD948KCtJTEmWcYizzhkGUekLO30AAApKD0AQAppjrf6/X6Z375923it2+2W+cKFC2V+9uxZ+wtjZLKMRZ5xyDKOqFna6QEAUlB6AIAUQh9vbW1tlfn+/ftDvWdlZaXMS0tLk14SY5JlLPKMQ5ZxZMjSTg8AkILSAwCkoPQAACmEfqbn48ePZf758+fA++7cuVPmJ0+etLomxiPLWOQZhyzjyJClnR4AIAWlBwBIIdTx1ocPHxrXq6urx9538+bNxvXhH1M7f/785BfGyGQZizzjkGUcGbO00wMApKD0AAApzPzx1jj/QXJhYaFx3ev1JrkkxiTLWOQZhyzjyJ6lnR4AIAWlBwBIQekBAFKY+Wd6Xr16VeZutzvUewb9WR7TJctY5BmHLOPInqWdHgAgBaUHAEhhJo+3fvz4Ueb19fWh3nPv3r0yX716deJrYjyyjEWeccgyDlkesNMDAKSg9AAAKczk8dbdu3fLvLu7O/C+69evl/nwD6RxesgyFnnGIcs4ZHnATg8AkILSAwCkoPQAACnM5DM9/X6/zH/7j5KPHz8u87lz51pdE+ORZSzyjEOWccjygJ0eACAFpQcASGEmjrcePnzYuK7rusz7+/sD33fjxo3W1sR4ZBmLPOOQZRyyHMxODwCQgtIDAKRwao+3Dv9A2qdPnxqvVVVV5rNnz5b50aNHjft6vV5Lq2MUsoxFnnHIMg5ZDsdODwCQgtIDAKSg9AAAKZzaZ3r29vbK/OvXr4H3XblypcyvX79udU2MR5axyDMOWcYhy+HY6QEAUlB6AIAUlB4AIAWlBwBIQekBAFJQegCAFE7tn6wvLi6W+egvv3779u2kl8M/kGUs8oxDlnHIcjh2egCAFJQeACCFU3u8dfny5TJ//fp1iivhX8kyFnnGIcs4ZDkcOz0AQApKDwCQgtIDAKSg9AAAKSg9AEAKSg8AkILSAwCkoPQAACkoPQBAClVd18PfXFX/dTqd7faWwzHm6rq+NOkPleXUyDMOWcYy8TxlOTUDsxyp9AAAzCrHWwBACkoPAJCC0gMApKD0AAApKD0AQApKDwCQgtIDAKSg9AAAKSg9AEAKSg8AkILSAwCkoPQAACkoPQBACkoPAJCC0gMApKD0AAApKD0AQApnRrn54sWL9fz8fEtL4ThbW1udfr9fTfpzZTkdm5ub/bquL036c+V58nw3Y2njuynL6fhbliOVnvn5+c7GxsZkVsVQrl271srnynI6qqrabuNz5XnyfDdjaeO7Kcvp+FuWjrcAgBSUHgAgBaUHAEhB6QEAUlB6AIAUlB4AIAWlBwBIQekBAFJQegCAFJQeACAFpQcASEHpAQBSUHoAgBSUHgAghTPTXsBp9uLFi8b18+fPy1zXdZm/fPnSuO/WrVutrovRyTIWecYhyzhmIUs7PQBACkoPAJCC460j3r17V+aXL182Xut2u2Xe398vc1VVra+L0ckyFnnGIcs4Zi1LOz0AQApKDwCQguOtI7a3t8v858+fKa6EfyXLWOQZhyzjmLUs7fQAACkoPQBACkoPAJCCZ3o6nc7nz5/L/ObNm4H3LS4ulnltba3MvV6vnYUxMlnGIs84ZBnHLGdppwcASEHpAQBSSHm89f3798b18vJymX///j3wfU+fPi3z3NzcxNfF6GQZizzjkGUckbK00wMApKD0AAApKD0AQAopn+l5//5943pnZ+fY+27fvt24fvDgQVtLYkyyjEWeccgyjkhZ2ukBAFJQegCAFNIcb/X7/TK/ffu28Vq32y3zhQsXyvzs2bP2F8bIZBmLPOOQZRxRs7TTAwCkoPQAACmEPt7a2toq8/3794d6z8rKSpmXlpYmvSTGJMtY5BmHLOPIkKWdHgAgBaUHAEhB6QEAUgj9TM/Hjx/L/PPnz4H33blzp8xPnjxpdU2MR5axyDMOWcaRIUs7PQBACkoPAJBCqOOtDx8+NK5XV1ePve/mzZuN68M/pnb+/PnJL4yRyTIWecYhyzgyZmmnBwBIQekBAFKY+eOtcf6D5MLCQuO61+tNckmMSZaxyDMOWcaRPUs7PQBACkoPAJCC0gMApDDzz/S8evWqzN1ud6j3DPqzPKZLlrHIMw5ZxpE9Szs9AEAKSg8AkMJMHm/9+PGjzOvr60O95969e2W+evXqxNfEeGQZizzjkGUcsjxgpwcASEHpAQBSmMnjrbt375Z5d3d34H3Xr18v8+EfSOP0kGUs8oxDlnHI8oCdHgAgBaUHAEhB6QEAUpjJZ3r6/X6Z//YfJR8/flzmc+fOtbomxiPLWOQZhyzjkOUBOz0AQApKDwCQwkwcbz18+LBxXdd1mff39we+78aNG62tifHIMhZ5xiHLOGQ5mJ0eACAFpQcASOHUHm8d/oG0T58+NV6rqqrMZ8+eLfOjR48a9/V6vZZWxyhkGYs845BlHLIcjp0eACAFpQcASEHpAQBSOLXP9Ozt7ZX5169fA++7cuVKmV+/ft3qmhiPLGORZxyyjEOWw7HTAwCkoPQAACkoPQBACkoPAJCC0gMApKD0AAApnNo/WV9cXCzz0V9+/fbt20kvh38gy1jkGYcs45DlcOz0AAApKD0AQAqn9njr8uXLZf769esUV8K/kmUs8oxDlnHIcjh2egCAFJQeACAFpQcASEHpAQBSUHoAgBSUHgAgBaUHAEhB6QEAUlB6AIAUqrquh7+5qv7rdDrb7S2HY8zVdX1p0h8qy6mRZxyyjGXiecpyagZmOVLpAQCYVY63AIAUlB4AIAWlBwBIQekBAFJQegCAFJQeACAFpQcASEHpAQBSUHoAgBSUHgAgBaUHAEhB6QEAUlB6AIAUlB4AIAWlBwBIQekBAFJQegCAFM6McvPFixfr+fn5lpbCcba2tjr9fr+a9OfKcjo2Nzf7dV1fmvTnyvPk+W7G0sZ3U5bT8bcsRyo98/PznY2NjcmsiqFcu3atlc+V5XRUVbXdxufK8+T5bsbSxndTltPxtywdbwEAKSg9AEAKSg8AkILSAwCkoPQAACkoPQBACkoPAJCC0gMApKD0AAApKD0AQApKDwCQgtIDAKSg9AAAKSg9AEAKSg8AkMKZaS/gNHvx4kXj+vnz52Wu67rMX758adx369atVtfF6GQZizzjkGUcs5ClnR4AIAWlBwBIwfHWEe/evSvzy5cvG691u90y7+/vl7mqqtbXxehkGYs845BlHLOWpZ0eACAFpQcASMHx1hHb29tl/vPnzxRXwr+SZSzyjEOWccxalnZ6AIAUlB4AIAWlBwBIwTM9nU7n8+fPZX7z5s3A+xYXF8u8trZW5l6v187CGJksY5FnHLKMY5aztNMDAKSg9AAAKaQ83vr+/Xvjenl5ucy/f/8e+L6nT5+WeW5ubuLrYnSyjEWeccgyjkhZ2ukBAFJQegCAFJQeACCFlM/0vH//vnG9s7Nz7H23b99uXD948KCtJTEmWcYizzhkGUekLO30AAApKD0AQAppjrf6/X6Z375923it2+2W+cKFC2V+9uxZ+wtjZLKMRZ5xyDKOqFna6QEAUlB6AIAUQh9vbW1tlfn+/ftDvWdlZaXMS0tLk14SY5JlLPKMQ5ZxZMjSTg8AkILSAwCkoPQAACmEfqbn48ePZf758+fA++7cuVPmJ0+etLomxiPLWOQZhyzjyJClnR4AIAWlBwBIIdTx1ocPHxrXq6urx9538+bNxvXhH1M7f/785BfGyGQZizzjkGUcGbO00wMApKD0AAApzPzx1jj/QXJhYaFx3ev1JrkkxiTLWOQZhyzjyJ6lnR4AIAWlBwBIQekBAFKY+Wd6Xr16VeZutzvUewb9WR7TJctY5BmHLOPInqWdHgAgBaUHAEhhJo+3fvz4Ueb19fWh3nPv3r0yX716deJrYjyyjEWeccgyDlkesNMDAKSg9AAAKczk8dbdu3fLvLu7O/C+69evl/nwD6RxesgyFnnGIcs4ZHnATg8AkILSAwCkoPQAACnM5DM9/X6/zH/7j5KPHz8u87lz51pdE+ORZSzyjEOWccjygJ0eACAFpQcASGEmjrcePnzYuK7rusz7+/sD33fjxo3W1sR4ZBmLPOOQZRyyHMxODwCQgtIDAKRwao+3Dv9A2qdPnxqvVVVV5rNnz5b50aNHjft6vV5Lq2MUsoxFnnHIMg5ZDsdODwCQgtIDAKSg9AAAKZzaZ3r29vbK/OvXr4H3XblypcyvX79udU2MR5axyDMOWcYhy+HY6QEAUlB6AIAUlB4AIAWlBwBIQekBAFJQegCAFE7tn6wvLi6W+egvv3779u2kl8M/kGUs8oxDlnHIcjh2egCAFJQeACCFU3u8dfny5TJ//fp1iivhX8kyFnnGIcs4ZDkcOz0AQApKDwCQgtIDAKSg9AAAKSg9AEAKSg8AkILSAwCkoPQAACkoPQBAClVd18PfXFX/dTqd7faWwzHm6rq+NOkPleXUyDMOWcYy8TxlOTUDsxyp9AAAzCrHWwBACkoPAJCC0gMApKD0AAApKD0AQApKDwCQgtIDAKSg9AAAKSg9AEAKSg8AkILSAwCkoPQAACkoPQBACkoPAJCC0gMApKD0AAApKD0AQApnRrn54sWL9fz8fEtL4ThbW1udfr9fTfpzZTkdm5ub/bquL036c+V58nw3Y2njuynL6fhbliOVnvn5+c7GxsZkVsVQrl271srnynI6qqrabuNz5XnyfDdjaeO7Kcvp+FuWjrcAgBSUHgAgBaUHAEhB6QEAUlB6AIAUlB4AIAWlBwBIQekBAFJQegCAFJQeACAFpQcASEHpAQBSUHoAgBSUHgAghTPTXsBp9uLFi8b18+fPy1zXdZm/fPnSuO/WrVutrovRyTIWecYhyzhmIUs7PQBACkoPAJCC460j3r17V+aXL182Xut2u2Xe398vc1VVra+L0ckyFnnGIcs4Zi1LOz0AQApKDwCQguOtI7a3t8v858+fKa6EfyXLWOQZhyzjmLUs7fQAACkoPQBACkoPAJCCZ3o6nc7nz5/L/ObNm4H3LS4ulnltba3MvV6vnYUxMlnGIs84ZBnHLGdppwcASEHpAQBSSHm89f3798b18vJymX///j3wfU+fPi3z3NzcxNfF6GQZizzjkGUckbK00wMApKD0AAApKD0AQAopn+l5//5943pnZ+fY+27fvt24fvDgQVtLYkyyjEWeccgyjkhZ2ukBAFJQegCAFNIcb/X7/TK/ffu28Vq32y3zhQsXyvzs2bP2F8bIZBmLPOOQZRxRs7TTAwCkoPQAACmEPt7a2toq8/3794d6z8rKSpmXlpYmvSTGJMtY5BmHLOPIkKWdHgAgBaUHAEhB6QEAUgj9TM/Hjx/L/PPnz4H33blzp8xPnjxpdU2MR5axyDMOWcaRIUs7PQBACkoPAJBCqOOtDx8+NK5XV1ePve/mzZuN68M/pnb+/PnJL4yRyTIWecYhyzgyZmmnBwBIQekBAFKY+eOtcf6D5MLCQuO61+tNckmMSZaxyDMOWcaRPUs7PQBACkoPAJCC0gMApDDzz/S8evWqzN1ud6j3DPqzPKZLlrHIMw5ZxpE9Szs9AEAKSg8AkMJMHm/9+PGjzOvr60O95969e2W+evXqxNfEeGQZizzjkGUcsjxgpwcASEHpAQBSmMnjrbt375Z5d3d34H3Xr18v8+EfSOP0kGUs8oxDlnHI8oCdHgAgBaUHAEhB6QEAUpjJZ3r6/X6Z//YfJR8/flzmc+fOtbomxiPLWOQZhyzjkOUBOz0AQApKDwCQwkwcbz18+LBxXdd1mff39we+78aNG62tifHIMhZ5xiHLOGQ5mJ0eACAFpQcASOHUHm8d/oG0T58+NV6rqqrMZ8+eLfOjR48a9/V6vZZWxyhkGYs845BlHLIcjp0eACAFpQcASEHpAQBSOLXP9Ozt7ZX5169fA++7cuVKmV+/ft3qmhiPLGORZxyyjEOWw7HTAwCkoPQAACkoPQBACkoPAJCC0gMApKD0AAApnNo/WV9cXCzz0V9+/fbt20kvh38gy1jkGYcs45DlcOz0AAApKD0AQAqn9njr8uXLZf769esUV8K/kmUs8oxDlnHIcjh2egCAFJQeACAFpQcASEHpAQBSUHoAgBSUHgAgBaUHAEhB6QEAUlB6AIAUqrquh7+5qv7rdDrb7S2HY8zVdX1p0h8qy6mRZxyyjGXiecpyagZmOVLpAQCYVY63AIAUlB4AIAWlBwBIQekBAFJQegCAFJQeACAFpQcASEHpAQBSUHoAgBT+D2JERZvO8SjjAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 720x720 with 25 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BP2D-QM5IXps",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "difference_between_images = tf.subtract(adverserial_training_images_batch_modified[3,:,:], train_images[3,:,:])"
      ],
      "execution_count": 41,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Rpp2qK0XIljQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "d2ad457c-3402-42ff-967b-cded3b5fedba"
      },
      "source": [
        "difference_between_images"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(28, 28), dtype=float32, numpy=\n",
              "array([[ 2.38417124e-06,  4.41849943e-06,  1.99540182e-06,\n",
              "        -1.59136744e-06,  6.60301794e-07,  2.59791955e-06,\n",
              "        -5.02706712e-07, -3.71205306e-06,  8.93047982e-05,\n",
              "         1.50695341e-05, -1.87625919e-04,  2.17212219e-05,\n",
              "        -8.01666465e-04, -2.38224660e-04, -2.28213830e-04,\n",
              "        -1.33787107e-04,  5.39062021e-05, -6.50173228e-04,\n",
              "        -5.95099846e-05, -2.16779692e-04,  8.96194324e-05,\n",
              "        -2.75785733e-05,  1.15444050e-06,  5.08311189e-07,\n",
              "         4.31255939e-05,  3.53241194e-05, -1.16408715e-04,\n",
              "        -3.01041964e-05],\n",
              "       [-3.17463855e-06, -2.38019584e-06, -2.66156280e-06,\n",
              "         1.85344129e-06,  2.05220113e-06,  4.45841124e-06,\n",
              "        -2.27090254e-06,  3.91269805e-06,  4.67858918e-05,\n",
              "         9.15504916e-06,  7.99205918e-06,  3.74566589e-04,\n",
              "        -1.11554179e-03, -7.20557466e-04, -5.32063597e-04,\n",
              "        -4.23532183e-04,  1.95943154e-04, -4.48371255e-04,\n",
              "         6.59516882e-05, -1.11770903e-04,  1.82604690e-05,\n",
              "         6.91184832e-05,  5.36575556e-07,  1.70600447e-06,\n",
              "         4.08960950e-05,  4.51386150e-05, -1.29323831e-04,\n",
              "        -1.34153852e-05],\n",
              "       [ 7.00870942e-06,  3.01974387e-06,  5.44697423e-06,\n",
              "         2.98194664e-06,  1.68010752e-06,  1.03279208e-05,\n",
              "         8.98197777e-06,  6.35378865e-06,  2.01631747e-06,\n",
              "        -4.75581032e-07,  2.31301819e-04,  5.16186003e-04,\n",
              "        -8.01059825e-04, -8.76053004e-04, -3.96553252e-04,\n",
              "        -1.05973589e-03,  1.05378160e-04,  2.15998822e-04,\n",
              "         2.93571880e-04,  1.43399913e-04,  6.56439579e-06,\n",
              "         3.82912003e-06,  1.50581184e-06,  5.22484527e-07,\n",
              "         1.50414162e-05,  2.76943028e-05, -6.58118297e-05,\n",
              "         5.81098720e-06],\n",
              "       [ 2.00995260e-06,  4.00741783e-06,  2.32527213e-06,\n",
              "         1.12625903e-05,  3.56662417e-06,  2.93343351e-06,\n",
              "        -3.34434162e-06,  5.44474551e-06,  4.05369110e-06,\n",
              "         1.20915329e-05,  1.27171108e-04,  4.18122945e-04,\n",
              "        -3.99805896e-04, -5.55583159e-04, -2.79914908e-04,\n",
              "        -2.18485016e-03, -4.17409989e-04, -1.61167045e-04,\n",
              "         4.26066836e-04,  3.24346503e-04, -1.81684447e-06,\n",
              "         2.20156267e-06,  1.90551248e-06,  4.61659283e-06,\n",
              "        -3.73783564e-06,  6.12533950e-06, -7.33068873e-06,\n",
              "         1.44166745e-06],\n",
              "       [ 2.81262101e-06,  6.40702297e-07,  4.49815298e-06,\n",
              "         7.15162685e-07, -3.12838210e-06,  2.42184228e-06,\n",
              "         4.07696916e-06, -8.37449352e-06, -6.47482648e-06,\n",
              "         5.11612279e-06, -4.37769103e-07,  2.23386291e-04,\n",
              "         1.45978483e-04, -5.57014428e-04,  6.75661868e-05,\n",
              "        -2.73448299e-03,  3.63539548e-05, -9.85990860e-04,\n",
              "         2.94902857e-05,  5.46028377e-06,  8.20943205e-06,\n",
              "         2.83335635e-06,  1.11118325e-07,  2.51560391e-06,\n",
              "        -4.26613497e-07,  2.81606253e-06,  1.58695148e-05,\n",
              "         3.31929095e-06],\n",
              "       [ 1.19797130e-06, -6.66334302e-07,  5.58595093e-06,\n",
              "         5.73346779e-06,  2.97215479e-06,  1.72403600e-06,\n",
              "         8.50920514e-06, -6.40235839e-07,  6.30604154e-06,\n",
              "         4.37912211e-04,  2.99544132e-04,  5.72863908e-04,\n",
              "         4.41140146e-04, -6.64344872e-04, -2.49445759e-04,\n",
              "        -1.53750461e-03, -3.91459143e-05, -3.77063516e-05,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00, -2.85413165e-07, -1.24799351e-06,\n",
              "        -1.80251948e-06,  4.44088755e-06,  1.05000086e-04,\n",
              "         7.23307021e-05],\n",
              "       [ 2.37362178e-06,  2.65527092e-06,  2.63218817e-06,\n",
              "         5.62258992e-06,  3.64711468e-07,  1.37697125e-05,\n",
              "         2.73852834e-06,  1.34895608e-05,  7.93680520e-05,\n",
              "         1.04652638e-04, -6.22274587e-04, -3.63983272e-04,\n",
              "        -9.39043472e-04, -1.52543804e-03, -1.07597793e-03,\n",
              "        -6.02335786e-04, -2.89486459e-04,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  2.12043687e-06, -7.64712968e-06,\n",
              "         2.85128772e-05,  1.53404544e-05, -3.65924498e-05,\n",
              "        -1.20691635e-04],\n",
              "       [-9.67274786e-07,  1.86738828e-06,  7.34009700e-06,\n",
              "         4.81542838e-06,  2.98418945e-06,  1.12582120e-05,\n",
              "         1.02937847e-05,  1.99495946e-04,  3.24716792e-04,\n",
              "        -8.30756675e-04, -8.64239817e-04, -4.16561030e-03,\n",
              "        -2.58763297e-03, -3.62666300e-03, -8.13948514e-04,\n",
              "        -9.26184875e-04, -7.72275325e-06,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  8.38332198e-06,  2.42503233e-06,\n",
              "         4.41558768e-05,  2.33141818e-05, -1.30982560e-04,\n",
              "         2.63034217e-05],\n",
              "       [ 1.72985915e-06, -1.13609310e-07, -2.31438158e-08,\n",
              "         2.75665201e-07, -2.37101449e-06, -7.61654519e-07,\n",
              "         5.64867196e-06,  3.92839778e-04,  4.12311318e-04,\n",
              "        -1.26641628e-03, -6.55540265e-04, -4.84503666e-03,\n",
              "        -3.09827318e-03, -3.67109198e-03, -7.69593171e-04,\n",
              "        -1.73414937e-05,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  1.00354046e-05,  3.25155111e-06,\n",
              "         2.83524078e-05,  2.10075650e-05, -5.00802344e-05,\n",
              "         1.45878876e-05],\n",
              "       [ 1.94237441e-06, -4.71599350e-07,  3.97145277e-06,\n",
              "         3.78539994e-06, -3.71054762e-06,  1.14364329e-05,\n",
              "         4.80952622e-06,  2.97001679e-04,  3.27934889e-04,\n",
              "        -1.01771322e-03, -3.74122668e-04, -3.08263046e-03,\n",
              "        -2.18932307e-03, -8.63476540e-04, -1.17845193e-03,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  5.19057551e-07,\n",
              "         2.42129363e-06,  6.62440107e-06,  4.73764021e-06,\n",
              "         5.35958975e-07,  1.89247567e-05,  8.12378858e-05,\n",
              "         9.31625709e-06],\n",
              "       [ 1.67928636e-06,  6.27118970e-06,  6.98363829e-06,\n",
              "        -7.42349641e-07,  1.33600531e-06,  4.68039752e-06,\n",
              "         4.11356632e-06,  8.06831449e-05,  2.91785313e-04,\n",
              "        -2.95798091e-04, -6.25444867e-04, -1.91875108e-04,\n",
              "        -1.56920590e-03,  2.87584640e-04, -7.97664234e-06,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00, -5.31195747e-06,  5.41572854e-06,\n",
              "         4.07670171e-07,  6.36173536e-06,  1.13181579e-06,\n",
              "         1.27302030e-06,  1.68931672e-06,  1.88682623e-06,\n",
              "         1.31966465e-06],\n",
              "       [-4.10986836e-08,  2.20498282e-06,  6.20452738e-06,\n",
              "         4.90261755e-06, -3.26577833e-06,  1.76586691e-05,\n",
              "         1.10379824e-06,  7.73110969e-06,  2.78985157e-04,\n",
              "         2.21156733e-04, -6.23242231e-04,  3.59108904e-04,\n",
              "        -1.21348479e-04,  4.52214881e-05,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  2.96380654e-06,  6.38684287e-06,\n",
              "         2.42000465e-06,  8.08613322e-07, -3.49142283e-06,\n",
              "        -4.20901233e-06,  7.79496804e-06,  5.48312619e-06,\n",
              "         4.41493376e-06],\n",
              "       [ 8.85585621e-07,  2.40054624e-06,  3.74529827e-06,\n",
              "         2.70834107e-06,  5.08895118e-06,  7.61228830e-06,\n",
              "        -2.17079233e-07, -5.21489255e-06,  1.95821820e-04,\n",
              "         2.18180881e-04, -3.03256675e-04, -1.37659163e-05,\n",
              "        -1.27031482e-04,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         8.05235140e-06,  6.90311890e-06,  7.90521062e-06,\n",
              "         3.62969672e-06,  4.50257949e-06,  6.67243876e-06,\n",
              "        -1.26565807e-08,  3.25985320e-06, -2.52601410e-07,\n",
              "         3.34916024e-07],\n",
              "       [ 5.10639347e-06,  2.00993281e-06,  1.83809746e-07,\n",
              "         8.72930832e-06, -2.49739628e-06,  1.80863317e-05,\n",
              "         3.08789276e-05,  1.63216551e-04, -1.62069009e-05,\n",
              "        -8.02456125e-05, -9.74669165e-05,  1.84580877e-05,\n",
              "         1.60317049e-05,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         1.13194423e-04,  4.11303336e-04,  5.71476412e-04,\n",
              "         4.88216523e-04,  4.42400196e-04,  2.23005598e-04,\n",
              "         6.41849256e-05,  1.12969181e-04,  2.88199517e-04,\n",
              "         1.74119123e-04],\n",
              "       [ 3.42486192e-06,  3.85049634e-06,  1.09621387e-05,\n",
              "         1.11680893e-05,  4.65607218e-06,  4.28551266e-06,\n",
              "         6.87813808e-05,  8.62512534e-05, -4.48467530e-04,\n",
              "        -3.00220862e-08, -2.38507870e-04,  1.49953103e-05,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  5.30545040e-05,\n",
              "        -1.44968071e-04,  3.05343099e-04,  2.60768749e-04,\n",
              "         3.26970512e-05, -2.40209320e-05,  3.31919364e-05,\n",
              "         4.24236641e-04,  8.31433790e-05, -2.14534175e-05,\n",
              "        -6.54494506e-05],\n",
              "       [ 5.50408004e-05,  5.00670103e-05,  1.42585224e-04,\n",
              "         9.25253116e-05,  3.02095887e-05,  1.34545335e-04,\n",
              "         2.23614872e-04, -1.64460653e-04, -5.09953708e-04,\n",
              "        -2.21061637e-04, -1.18221797e-04,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00, -9.78325959e-04, -1.64715585e-03,\n",
              "        -3.00160935e-03, -1.52391323e-03, -1.86619372e-03,\n",
              "        -1.46700174e-03, -1.20641617e-03, -5.26940741e-04,\n",
              "        -3.71593691e-04, -4.06461826e-04, -3.67998931e-04,\n",
              "        -3.96303774e-04],\n",
              "       [ 5.69409858e-05, -1.11185305e-04,  1.59465184e-04,\n",
              "         1.61666539e-04,  1.76486719e-04,  2.15120424e-04,\n",
              "         2.02591880e-04, -1.97209039e-04, -5.19506611e-05,\n",
              "        -1.77999507e-04, -9.62848662e-06,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00, -2.55599921e-03, -2.49400083e-03,\n",
              "        -5.60127245e-03, -3.28078959e-03, -4.29602433e-03,\n",
              "        -2.04797322e-03, -1.60152558e-03, -5.11893188e-04,\n",
              "        -4.95753302e-05, -9.39669902e-04, -6.07438502e-04,\n",
              "        -2.61430774e-04],\n",
              "       [-3.62370483e-04, -3.00618649e-05, -7.13977090e-04,\n",
              "        -2.39910543e-04, -5.18051325e-04, -4.55749192e-04,\n",
              "        -3.13611759e-04,  7.05073035e-05,  2.44231342e-04,\n",
              "         6.47708148e-05,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "        -6.53022667e-04, -4.07554489e-03, -2.67942599e-03,\n",
              "        -6.03577727e-03, -1.76185940e-03, -3.67341423e-03,\n",
              "        -1.01177045e-03, -1.60376460e-03, -3.47555178e-05,\n",
              "        -2.35956599e-04, -8.06799144e-05, -2.54992192e-04,\n",
              "        -2.43842573e-04],\n",
              "       [-4.48930514e-04,  3.03005683e-04, -1.22246577e-03,\n",
              "         1.10855712e-04, -9.90218716e-04, -1.86791900e-03,\n",
              "        -4.34217771e-04, -3.28292692e-04,  1.15284194e-04,\n",
              "         2.98747938e-04,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "        -8.84149864e-04, -3.50971427e-03, -2.56881281e-03,\n",
              "        -4.31421772e-03, -1.00473326e-03, -2.19287118e-03,\n",
              "        -2.14491665e-04, -1.12340436e-03,  1.73033026e-04,\n",
              "         2.31637052e-04, -1.20832592e-04, -1.22956379e-04,\n",
              "        -1.81186202e-04],\n",
              "       [-1.64751909e-04,  2.87840696e-04, -4.03846236e-04,\n",
              "         1.44586433e-04, -5.01978269e-04, -8.98983912e-04,\n",
              "        -3.92110320e-04, -5.18206725e-05, -7.19262825e-05,\n",
              "         3.20561048e-06,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  1.23798251e-04,\n",
              "        -4.60483512e-04, -2.47955183e-03, -1.75033836e-03,\n",
              "        -2.53093429e-03, -6.05396519e-04, -9.33935691e-04,\n",
              "        -1.95168061e-04, -1.09527144e-04,  3.52344068e-06,\n",
              "         1.75404304e-04, -4.59152885e-04, -3.95127194e-04,\n",
              "        -1.63109464e-04],\n",
              "       [ 1.10846653e-04, -6.65910557e-06, -8.37201078e-05,\n",
              "         1.65364094e-04, -2.95329286e-04,  1.21003119e-04,\n",
              "        -9.54475472e-05,  1.86296966e-04, -1.58345538e-05,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  3.10491916e-04,\n",
              "        -4.56044771e-04, -1.48589839e-03, -2.31732009e-03,\n",
              "        -1.39376509e-03, -1.29734946e-03, -7.05384242e-04,\n",
              "        -4.76177753e-04,  1.75909241e-04,  2.61098990e-04,\n",
              "        -1.25420265e-04, -5.24618139e-04, -4.42423247e-04,\n",
              "        -1.75947920e-04],\n",
              "       [-2.36422366e-05,  5.89136871e-05, -1.29413020e-04,\n",
              "         7.16576906e-05, -7.27907945e-07,  2.15303022e-04,\n",
              "         8.87796996e-05,  1.36610606e-06,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  4.51731088e-04,\n",
              "        -7.64629338e-04, -8.38970882e-04, -1.96052319e-03,\n",
              "        -1.50229060e-03, -9.20393039e-04, -6.56890101e-04,\n",
              "        -1.95896471e-04, -1.40476637e-04, -1.23455320e-04,\n",
              "        -3.18636827e-04, -2.09625039e-04, -1.81482246e-04,\n",
              "        -1.23178339e-04],\n",
              "       [-4.23105375e-05, -6.01352394e-06, -4.69012011e-05,\n",
              "         6.08010014e-05,  9.99401964e-05,  8.34095772e-06,\n",
              "         2.68804774e-06,  3.10534961e-06,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "        -4.97107976e-04, -6.04832487e-04, -2.80714274e-04,\n",
              "        -7.50359788e-04, -2.35190368e-04, -9.36042226e-04,\n",
              "        -4.36913106e-04, -3.07536422e-04, -1.04465878e-04,\n",
              "        -4.98887093e-04, -2.93588964e-04, -1.97756366e-04,\n",
              "        -8.22307193e-04, -6.55911397e-04, -5.42560709e-04,\n",
              "        -1.73211738e-04],\n",
              "       [ 1.88800902e-06, -3.35108125e-05,  5.09826750e-05,\n",
              "         2.86207160e-06,  3.19537753e-06,  4.87654745e-07,\n",
              "        -2.48373340e-06,  3.24801431e-06,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "        -3.53955635e-04, -7.22881057e-04, -1.03670161e-03,\n",
              "        -2.53574661e-04,  3.67402281e-05, -3.96821997e-04,\n",
              "         1.97281493e-04, -5.84936861e-05,  2.22780800e-04,\n",
              "        -6.54771749e-04, -1.20657016e-04, -7.96479813e-04,\n",
              "        -5.10215643e-04, -1.05444854e-03, -7.95793661e-04,\n",
              "        -4.69818595e-04],\n",
              "       [ 1.12712787e-05, -3.72199270e-06,  4.79196206e-06,\n",
              "         5.84631516e-06,  4.64119103e-06,  1.98800331e-06,\n",
              "        -1.78544985e-06, -5.79592097e-07,  0.00000000e+00,\n",
              "         0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
              "         2.81640823e-04, -8.42214446e-04,  4.15268529e-04,\n",
              "        -1.94100692e-04,  3.33770149e-04,  7.32619810e-05,\n",
              "         1.33203444e-04,  1.32799149e-04, -4.47149723e-05,\n",
              "        -2.10129045e-04, -9.47529315e-06, -4.54673864e-04,\n",
              "        -7.65038247e-04, -6.06216723e-04, -6.55140437e-04,\n",
              "        -1.62318000e-04],\n",
              "       [-1.30099215e-06,  6.99313068e-06,  3.97853728e-06,\n",
              "         2.92679783e-06,  6.68736629e-06,  1.69827285e-06,\n",
              "         1.94512631e-06,  3.12246698e-06,  2.94960114e-06,\n",
              "         2.52167956e-06, -3.76422554e-06,  2.21465916e-06,\n",
              "         4.17107774e-04, -1.75837529e-04,  1.72617059e-04,\n",
              "         3.24445195e-04,  1.03373250e-05,  7.34305604e-06,\n",
              "         6.55304029e-06,  5.55773840e-05,  3.99662931e-05,\n",
              "         1.37947340e-04,  1.43450579e-05, -3.09389070e-05,\n",
              "        -2.81277491e-04, -2.30680322e-04, -1.96822759e-04,\n",
              "        -3.75094351e-06],\n",
              "       [-1.74270781e-06,  1.06685502e-06,  2.09919199e-06,\n",
              "         1.94186828e-06,  8.82281199e-07,  1.72172463e-06,\n",
              "         2.16045692e-06,  8.07519336e-06,  3.44243381e-06,\n",
              "        -3.29629552e-06,  3.46647494e-06, -1.14440422e-06,\n",
              "        -2.41458338e-06,  1.38419928e-04,  6.07661013e-06,\n",
              "         7.57186399e-06,  5.23045219e-06,  1.37681172e-06,\n",
              "        -1.13375145e-05,  4.99932230e-06,  7.49597821e-05,\n",
              "         1.29517503e-04, -5.60939088e-06, -1.22831061e-05,\n",
              "        -3.00862575e-05,  1.08246684e-04,  8.73652243e-05,\n",
              "         4.89418999e-05],\n",
              "       [-5.70338386e-07, -2.18968194e-06,  1.92323819e-06,\n",
              "         8.35033177e-07,  1.54741781e-06,  1.30238436e-06,\n",
              "         3.50786263e-06,  1.28244767e-06,  7.13609552e-06,\n",
              "         6.75239153e-06,  4.43264753e-06,  1.91155414e-06,\n",
              "         8.50822232e-07,  1.97168310e-06,  3.03524166e-06,\n",
              "         4.94222331e-06,  2.59726085e-06, -1.73247972e-05,\n",
              "        -1.80550649e-06, -1.87676615e-05,  2.91953211e-06,\n",
              "         8.46762838e-09,  5.37648191e-07,  6.42320229e-05,\n",
              "         2.41096532e-05,  4.04855928e-05,  4.83413824e-05,\n",
              "         5.06310062e-06]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "APOUFKGLAWoK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "5fbed36c-89b2-4143-96a2-ade38f6db5d4"
      },
      "source": [
        "plt.imshow(adverserial_training_images_batch_modified[3,:,:])"
      ],
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7feb831b6358>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 44
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-jIhEaS-IN-G",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "e4851fcb-f11e-461b-d497-915176f7fd78"
      },
      "source": [
        "plt.imshow(train_images[3,:,:])"
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7feb832b1f98>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 45
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UqYzgnpmAWiV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4d76gAJaAWfH",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WoVfiDfaAWbw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TXCWHuShAWaR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U7awl4RmAWTr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}