Lesson 5-1 MNIST Digit Recognition Accuracy 98%+.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.examples.tutorials.mnist import input_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting MNIST_data/train-images-idx3-ubyte.gz\n",
      "Extracting MNIST_data/train-labels-idx1-ubyte.gz\n",
      "Extracting MNIST_data/t10k-images-idx3-ubyte.gz\n",
      "Extracting MNIST_data/t10k-labels-idx1-ubyte.gz\n",
      "Iter0, Testing Accuracy: 0.9501, Learning Rate: 0.001\n",
      "Iter1, Testing Accuracy: 0.9633, Learning Rate: 0.00095\n",
      "Iter2, Testing Accuracy: 0.9681, Learning Rate: 0.0009025\n",
      "Iter3, Testing Accuracy: 0.9723, Learning Rate: 0.000857375\n",
      "Iter4, Testing Accuracy: 0.9704, Learning Rate: 0.00081450626\n",
      "Iter5, Testing Accuracy: 0.9752, Learning Rate: 0.0007737809\n",
      "Iter6, Testing Accuracy: 0.9761, Learning Rate: 0.0007350919\n",
      "Iter7, Testing Accuracy: 0.9766, Learning Rate: 0.0006983373\n",
      "Iter8, Testing Accuracy: 0.9775, Learning Rate: 0.0006634204\n",
      "Iter9, Testing Accuracy: 0.9769, Learning Rate: 0.0006302494\n",
      "Iter10, Testing Accuracy: 0.9791, Learning Rate: 0.0005987369\n",
      "Iter11, Testing Accuracy: 0.9787, Learning Rate: 0.0005688001\n",
      "Iter12, Testing Accuracy: 0.9785, Learning Rate: 0.0005403601\n",
      "Iter13, Testing Accuracy: 0.9798, Learning Rate: 0.0005133421\n",
      "Iter14, Testing Accuracy: 0.9801, Learning Rate: 0.000487675\n",
      "Iter15, Testing Accuracy: 0.981, Learning Rate: 0.00046329122\n",
      "Iter16, Testing Accuracy: 0.9824, Learning Rate: 0.00044012666\n",
      "Iter17, Testing Accuracy: 0.9813, Learning Rate: 0.00041812033\n",
      "Iter18, Testing Accuracy: 0.9814, Learning Rate: 0.00039721432\n",
      "Iter19, Testing Accuracy: 0.981, Learning Rate: 0.0003773536\n",
      "Iter20, Testing Accuracy: 0.9798, Learning Rate: 0.00035848594\n",
      "Iter21, Testing Accuracy: 0.9809, Learning Rate: 0.00034056162\n",
      "Iter22, Testing Accuracy: 0.9819, Learning Rate: 0.00032353355\n",
      "Iter23, Testing Accuracy: 0.9814, Learning Rate: 0.00030735688\n",
      "Iter24, Testing Accuracy: 0.9811, Learning Rate: 0.000291989\n",
      "Iter25, Testing Accuracy: 0.9819, Learning Rate: 0.00027738957\n",
      "Iter26, Testing Accuracy: 0.9834, Learning Rate: 0.0002635201\n",
      "Iter27, Testing Accuracy: 0.9821, Learning Rate: 0.00025034408\n",
      "Iter28, Testing Accuracy: 0.9821, Learning Rate: 0.00023782688\n",
      "Iter29, Testing Accuracy: 0.9806, Learning Rate: 0.00022593554\n",
      "Iter30, Testing Accuracy: 0.9819, Learning Rate: 0.00021463877\n",
      "Iter31, Testing Accuracy: 0.9807, Learning Rate: 0.00020390682\n",
      "Iter32, Testing Accuracy: 0.9809, Learning Rate: 0.00019371149\n",
      "Iter33, Testing Accuracy: 0.9811, Learning Rate: 0.0001840259\n",
      "Iter34, Testing Accuracy: 0.9823, Learning Rate: 0.00017482461\n",
      "Iter35, Testing Accuracy: 0.9819, Learning Rate: 0.00016608338\n",
      "Iter36, Testing Accuracy: 0.9815, Learning Rate: 0.00015777921\n",
      "Iter37, Testing Accuracy: 0.9806, Learning Rate: 0.00014989026\n",
      "Iter38, Testing Accuracy: 0.9822, Learning Rate: 0.00014239574\n",
      "Iter39, Testing Accuracy: 0.9822, Learning Rate: 0.00013527596\n",
      "Iter40, Testing Accuracy: 0.9824, Learning Rate: 0.00012851215\n",
      "Iter41, Testing Accuracy: 0.9822, Learning Rate: 0.00012208655\n",
      "Iter42, Testing Accuracy: 0.9823, Learning Rate: 0.00011598222\n",
      "Iter43, Testing Accuracy: 0.9819, Learning Rate: 0.00011018311\n",
      "Iter44, Testing Accuracy: 0.9821, Learning Rate: 0.000104673956\n",
      "Iter45, Testing Accuracy: 0.9822, Learning Rate: 9.944026e-05\n",
      "Iter46, Testing Accuracy: 0.9823, Learning Rate: 9.446825e-05\n",
      "Iter47, Testing Accuracy: 0.982, Learning Rate: 8.974483e-05\n",
      "Iter48, Testing Accuracy: 0.9825, Learning Rate: 8.525759e-05\n",
      "Iter49, Testing Accuracy: 0.9824, Learning Rate: 8.099471e-05\n",
      "Iter50, Testing Accuracy: 0.9823, Learning Rate: 7.6944976e-05\n"
     ]
    }
   ],
   "source": [
    "#load dataset\n",
    "mnist = input_data.read_data_sets(\"MNIST_data\",one_hot=True)\n",
    "\n",
    "#define batch size\n",
    "batch_size = 100\n",
    "#calculate number of batches\n",
    "n_batch = mnist.train.num_examples // batch_size\n",
    "\n",
    "#define placeholders\n",
    "x = tf.placeholder(tf.float32, [None,784])\n",
    "y = tf.placeholder(tf.float32, [None,10])\n",
    "keep_prob=tf.placeholder(tf.float32)\n",
    "lr = tf.Variable(0.001,dtype=tf.float32)\n",
    "\n",
    "#create simple NeuroNet\n",
    "W1 = tf.Variable(tf.truncated_normal([784,500],stddev=0.1))\n",
    "b1 = tf.Variable(tf.zeros([500])+0.1)\n",
    "L1 = tf.nn.tanh(tf.matmul(x,W1)+b1)\n",
    "L1_drop = tf.nn.dropout(L1,keep_prob)\n",
    "\n",
    "W2 = tf.Variable(tf.truncated_normal([500,300],stddev=0.1))\n",
    "b2 = tf.Variable(tf.zeros([300])+0.1)\n",
    "L2 = tf.nn.tanh(tf.matmul(L1_drop,W2)+b2)\n",
    "L2_drop = tf.nn.dropout(L2,keep_prob)\n",
    "\n",
    "W3 = tf.Variable(tf.truncated_normal([300,10],stddev=0.1))\n",
    "b3 = tf.Variable(tf.zeros([10])+0.1)\n",
    "prediction = tf.nn.softmax(tf.matmul(L2_drop,W3)+b3)\n",
    "\n",
    "#cost function\n",
    "loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y,logits=prediction))\n",
    "#train with gradient descent\n",
    "train_step = tf.train.AdamOptimizer(lr).minimize(loss)\n",
    "\n",
    "#initialize variables\n",
    "init = tf.global_variables_initializer()\n",
    "\n",
    "#find accuracy of trained model\n",
    "correct_prediction = tf.equal(tf.argmax(y,1),tf.argmax(prediction,1))   #convert a list of booleans into a single boolean value\n",
    "accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    sess.run(init)\n",
    "    for epoch in range(51):\n",
    "        sess.run(tf.assign(lr, 0.001 * (0.95 ** epoch)))\n",
    "        for batch in range(n_batch):\n",
    "            batch_xs,batch_ys = mnist.train.next_batch(batch_size)\n",
    "            sess.run(train_step,feed_dict={x:batch_xs,y:batch_ys,keep_prob:1.0})\n",
    "        \n",
    "        test_acc = sess.run(accuracy,feed_dict={x:mnist.test.images,y:mnist.test.labels,keep_prob:1.0})\n",
    "        print(\"Iter\" + str(epoch) + \", Testing Accuracy: \" + str(test_acc) + \", Learning Rate: \" + str(sess.run(lr)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}