2022/2023 Team Projects

PlayFitt Recommender System

TPMs: Ben Bates, Richard Wills
Core Members: Will Erf, Kristof Sochan, Nolan White-Roy, Ross Cleary
Our objective is to help people be more active! We developed a recommender system for the PlayFitt fitness app using a contextual bandit algorithm. This takes into account information about the users as context, and makes decisions about rep counts and reward values to suggest. Notably, this approach enables online learning.

In collaboration with Intellisports

Solar Photovoltaic Output Prediction

TPMs: Areel Khan, Carter Demars,
Core Members: Anupra Chandran, Bill Bai, Mihir Gupta, Raihan Abdul Vaheed
We set out to develop deep learning models that predict solar photovoltaic output. These forecasts on solar energy production can drive better energy decisions, massively reducing carbon emmissions produced by power grids.


In collaboration with Open Climate Fix

Prostate Cancer Prediction With Correlated Diffusion Imaging

TPMs: Hargun Mujral, Jarett Dewbury
Core Members: Adam Lam, Alex Shao, Michelle Watson, Praajna Baragur
The application of machine learning to medical images has led to impressive advancement in cancer diagnostics. Our objective is to develop a baseline of deep learning models to detect the presence of prostate cancer within a novel Correlated Diffusion Imaging (CDI) dataset.

In collaboration with Dr. Alexander Wong & Hayden Gunraj

Deep Learning Framework Comparison

TPMs: Anusha Raisinghani, Trevor Yu
Core Members: Adish Shah, Ethan Gabriel, Musaab Siddiqui, Urban Pistek, Yask Pokra
The goal of our project is to benchmark performance of several deep learning frameworks, including TensorFlow, PyTorch, Jax, MxNet, Flux.jl, and KNet.jl. We build models from scratch in each framework and test them on common datasets.

Reinforcement Learning Chess Engine

TPMs: Amya Singhal, Thomas Fortin
Core Members: Alex He, Alexander Hutchinson, Ishaan Patel, Shivam Jindal
We are developing an artificial intelligence chess engine based on the work of Deep Mind on their chess engine‚ĒĀAlpha-Zero. We are also iteratively testing and modifying the model to improve performance on hardware with much more limited processing power than what was available to Deep Mind when creating Alpha-Zero.

Stable Diffused Adversarial Attacks

TPMs: Andy Wu, Dhrumil Patel, Rayaq Siddiqui
Core Members: Eric Sheen, Harsh Patel, Iliad Shaghaghi, Nicole Jin, Ryan Shen
We are exploring pre-trained computer vision models (i.e. MobileNet_v1) with the goal to exploit vulnerabilities through various adversarial attacks. Our work aims to develop an architecture that automates adversarial attack image generation in model misclassification.

Our Partners