top of page

Modeling the ESRB

This used binary and multiclass sequential neural networks to model the expected ESRB rating of video games using features related to game content.

LSTMs for BBC articles and Nursery Rhymes

This was a two-pronged foray into LSTMs and their capabilities in NLP. The BBC project was an attempt to categorize articles using a bidirectional LSTM, whereas the Nursery Rhymes project was an attempt to train a generative model that could write nursery rhymes using prompts.

Deep-Q Learning

An exploration of Deep-Q learning and it's applications for modeling game-playing agents; this project focused on training models to play Atari Breakout, Super Mario Bros., and Geometry Dash.

Classification and Augmentation with MNIST

This explored both classification with convolutional neural networks as well as generative modeling with a C-VAE and a C-GAN using the Handwritten Numbers from the MNIST dataset.

bottom of page