This repository contains my solutions [completed] to the assignments of the CS231n course. A big thank you to the faculty of the course at Stanford for making the resource materials available.
- K-Nearest Neighbours
- Support Vector Machines
- Softmax
- Two-Layer Neural Network
(TODO: slight changes to calculation of deltas in backprop for softmax layer) - Features
- Fully-connected Nets
- Batchnormalisation
- Dropout
- Convolutional Neural Network
(TODO: Improve classification accuracy on CIFAR-10 above 65%) - Implement parametric ReLU
- Analyse effect of Adam vs Adam+LR decay
- Different ConvNet Architectures (particularly:
[conv - relu - conv - relu - pool]xM - [affine]xN - [softmax]
(TODO: debug the batchnorm implementation of this architecture)
- Image captioning with vanilla RNNs
- Image captioning with LSTMs
- Image Gradients
- Image Generation