My Projects

Clinical Machine Learning based on Cardiorespiratory Models and Simulation

This project aims to develop a learning model and simulation for integrating big health into support systems in a systematic manner, with application in Cardiorespiratory critical care. this project was initiated by CHU- Saints-Justine Research Center and working under the supervision of Professor Irina Rish. The first version of this work which is studying the Generative Foundation Models for ICU Vital Signs Prediction: A Time-Series Evaluation is submitted to the ML4LMS workshop at ICML 2024.

Lag-Llama: Towards Foundation Models for Probabilistic Time Series Forecasting

Foundation models have revolutionized machine learning, but time series forecasting has lagged behind. Introducing Lag-Llama, a versatile foundation model for univariate probabilistic time series forecasting using a decoder-only transformer with lags as covariates. Pretrained on diverse time series data, it shows excellent zero-shot generalization and achieves state-of-the-art performance when fine-tuned on small, unseen datasets. Lag-Llama, an open-source model, marks a significant advancement in time series forecasting and is the first of its kind accepted at the RO-FoMo NeurIPS 2023 workshop.

ProjectX program 

This is a 5  long ML research for undergrads to participate in and I'm working as a mentor. this is the main website of this project that is done with the Vector Institute in Toronto, Ontario. This project is from September 2021 until February 2022.