Deep latent variable model for relational learning and variational inference with backpropagation(all the derivations can be found in the folder)

Convolutional Neural Networks (implemented from scratch) for text classification(all the derivations can be found in the folder)

Neural Collaborative Filtering (https://arxiv.org/abs/1708.05031) with implicit responses. As I expected, it doesn't perform well.

Character level RNN for name classification is implemented with Julia language deep learning package Flux. Almost same implementation with https://pytorch.org/tutorials/intermediate/char_rnn_classification_tutorial.html

Lorenz system dynamics are learned with LSTM networks.