Projects

Graph-Based Weather Forecasting

Hierarchical GNNs using ERA5 data for the French Northwest coast.

Deep Reinforcement Learning

Recap of the key contributions of Matteo Hessel [2018], detailing the integration of distinct improvements in deep reinforcement learning. We provide theoretical insights into the mathematical foundations of these techniques and their interactions and provides ablation studies on Atari benchmark.

Generative Learning Trilemma

Exploring the trade-offs between sample quality, diversity, and computational efficiency in generative models.

Bayesian Analysis for DL

Bayesian modeling on diabetes and energy datasets inspired by Bayesian Deep Learning.

Recommendation Systems

Exploring machine learning methods, from basics to deep learning, for personalized recommendations using the 2023 Amazon Reviews dataset.

LLM Jailbreak – QROA

Black-box Query-Response Optimization Attack on LLMs. In collab. with Capgemini Invent.

Time Series Analysis

Monthly evolution of the manufacturing of medical and dental instruments and supplies from January 1990 to January 2020.

Eurozone Sentiment Analysis

Predicting Eurobarometer results using media sentiment. Banque de France research.

Denoising Acoustic Signals for Classification

Paper on denoising acoustic signals framework to improve classification performance on marine mammal vocalizations. Submitted to a Neurips2025 conference workshop.