Introduction

I've been taking Stanford course CS231n: Convolutional Neural Networks for Visual Recognition while studying at GIKI. Here I have gathered my notes and solutions to assignments. The course lectures were recorded in Spring 2017, but the assignments are from Spring 2024.

Notes

I have organized the notes by assignment, including related helpers that were instrumental in solving the assignments. The notes also feature handwritten derivations and hand-drawn diagrams of backpropagation.

• Assignment 1
• Assignment 2
• Assignment 3

Assignment Solutions

Assignment 1

You can get starter code from here

• Q1: k-Nearest Neighbor classifier.
• Q2: Training a Support Vector Machine.
• Q3: Implement a Softmax classifier.
• Q4: Two-Layer Neural Network.
• Q5: Higher Level Representations: Image Features.

Assignment 2

You can get starter code from here

• Q1: Multi-Layer Fully Connected Neural Networks.
• Q2: Batch Normalization.
• Q3: Dropout.
• Q4: Convolutional Neural Networks.
• Q5: PyTorch.

Assignment 3

You can get starter code from here

• Q1: Image Captioning with Vanilla RNNs.
• Q2: Image Captioning with Transformers.
• Q3: Generative Adversarial Networks.
• Q4: Self-Supervised Learning for Image Classification.
• Q5: Network Visualization: Saliency Maps, Class Visualization, and Fooling Images.
• Q6: Image Captioning with LSTMs.
• Q7: Style Transfer

Helpful Resources

• Backpropagation for Batch Normalization
  • Helped solidify my understanding of backpropagation
• StatQuest
  • Assisted in understanding concepts that were not fully clear from the course lectures
• ResNet
  • Helped solidfy my understanding of Residual Blocks and how they help in training deeper networks
• Transformers
  • Helped in understanding the transformer architecture.