Projects

I have performed multiple small projects spanning most areas of NLP. Some topics are word embedding, word counting, word correction and n-gram models. I am familiar with multiple NLP libraries such as NLTK and spaCy. Click here to see some of my NLP projects!

This was my attempt at writing the attention mechanism in https://arxiv.org/abs/1706.03762 from scratch to become familiar with it. The code can be found here .

This is an implementation of a sentiment analysis model for online reviews with huggingface and bert-base-multilingual-uncased-sentiment. The code can be found here .

This is an implementation of a bert model from huggingface for question-answering task. The code can be found here .

The purpose of this project was to perform a time series analysis. This was done with a recurrent neural network from the Keras library. The code can be found here .

This is my implementation of the cycleGAN from https://arxiv.org/abs/1703.10593 . The code can be found here. I also had to present this project for the class ift6164 at MILA with some slides.

This project is an implementation of an imitation from observation algorithm. The algorithm trains an agent to learn to imitate an expert performing some tasks, by using videos of this expert. We use the DrQv2 from https://arxiv.org/abs/2107.09645 . This was a project for class ift6163 and we had to do a presentation. The code can be found here. We tried to apply this algorithm to a real-life robot hand, check out this video.

For this project, we had to gather data from the NHL data API and build a full pipeline from tidying the data with pandas library and json files to visualization of player performance based on machine learning algorithms. This was a project for the class ift6758 and some code can be found here.

For the class ift6135, we had to implement some parts of different deep learning architectures such as CNN, LSTM, ViT, VAE, WGAN and SIMSIAM. The code with solutions for these projects can be found here.

In this project, we implemented a gaussian mixture model to analyze features in a dataset from asthma patients in order to determine whether they were in states “baseline”, “exacerbation” or “follow-up”. This was a project for ift6269 and our final report can be found here.