Custom machine learning implementations of various ML architectures for knowledge using only basic libraries.
Currently includes:
- Simple Linear Regression
- Fully Connected Neural Network
- Convolutional Neural Network (dynamic layers and activation function support)
Generative Adversarial Network (GAN) implementation with PyTorch (could be implemented with a few tweaks to the aforementioned CNN implementation, but this is much easier)
Examples:
MNIST:
CelebA (faces)
