CS231n assignments

Course links

• CS231n: Convolutional Neural Networks for Visual Recognition (Spring 2017), syllabus, 2016 video, 2017 videos, 2017 slides, 2021 slides
• 2019 Syllabus and Assignments
• Github: cs231n/cs231n.github.io
  • Andrei’s fork with solutions: Bitdribble/cs231n

Andrei’s assignments

Github: bitdribble/cs231n

• Assignment 1: assignments/2021/assignment1
  • knn.ipynb
  • svm.ipynb
  • softmax.ipynb
  • fc_net.py: implementation of fully connected net
  • k_nearest_neighbor.py
  • linear_classifier.py
  • softmax.py
  • layers.py implementations for:
    • affine_forward(), affine_backward()
    • relu_forward(), relu_backward()
    • softmax_loss()
  • layer_utils.py implementations for:
    • affine_relu_forward(), affine_relu_backward()
  • features.ipynb
• Assignment 2: assignments/2021/assignment2
  • FullyConnectedNets.ipynb
  • BatchNormalization.ipynb
  • Dropout.ipynb
  • optim.py: RMSProp, Adam optimizers
  • layers.py implementations for:
    • affine_forward(), affine_backward()
    • relu_forward(), relu_backward()
    • batchnorm_forward(), batchnorm_backward(). The backward pass is a good example of gradient descent in the compute graph.
    • layernorm_forward(), layernorm_backward()
    • softmax_loss()
  • layer_utils.py implementations for:
    • affine_relu_forward(), affine_relu_backward()
    • affine_bn_relu_forward(), affine_bn_relu_backward()
    • affine_ln_relu_forward(), affine_ln_relu_backward()
    • dropout_forward(), dropout_backward()
  • fc_net.py: implementation of fully connected net, with support for:
    • batch normalization
    • layer normalization
    • dropout
• Assignment 3: assignments/2021/assignment2

Other forks with solutions:

• jingxinfu/cs231n
• haofeixu/cs231n
• https://www.codeleading.com/article/83371501641/
• https://bywmm.github.io/categories/cs231n/

Resources

• Batch Normalization, gradient descent backward pass:
  • Flair of Machine Learning blog: Understanding the backward pass through Batch Normalization Layer, Frederik Kratzert (2016)
  • Kevin Zakka’s blog: Deriving the Gradient for the Backward Pass of Batch Normalization (2016)
• Normalization
  • Melissa Mozifian: Normalization Techniques - Neural Networks (2018)