CS231n

Spring 2019 assignments of CS231n: Convolutional Neural Networks for Visual Recognition
Notes in Russian with additional explanations and links.

Assignment #1. Image Classification on CIFAR-10: kNN, SVM, Softmax, Neural Network

Q1: k-Nearest Neighbor classifier. Test accuracy: 28.2%

Q2: Training a Support Vector Machine. Test accuracy: 37.6%

Q3: Implement a Softmax classifier. Test accuracy: 34%

Q4: Two-Layer Neural Network. Test accuracy: 54%

Q5: Higher Level Representations: Image Features. Test accuracy: 58.8%

Assignment #2: Fully-Connected Nets, Batch Normalization, Dropout, Convolutional Nets

Q1: Fully-connected Neural Network. Test accuracy: 59.5%

Q2: Batch Normalization

Q3: Dropout

Q4: Convolutional Networks including Group Normalization.
Extra: Search by parameter values

Q5: PyTorch. AlexNet.. Test accuracy: 74.38%
Extra: PyTorch. Simplified AlexNet.. Test accuracy: 78.20%
Extra2: PyTorch. Simplified ResNet.. Test accuracy: 73.99%
Extra3: PyTorch. VGG and ResNet models.