Projects

In this sections you can retrive some archived work.

Intern at Liris

I'm doing research on link prediction with graph contrastive learning methods.
Supervisor: Rémy Cazabet and Mathieu Lefort
Abstract: > Recently, instance discrimination models have emerged as a major solution for self-supervised learning. Having already demonstrated its effectiveness in the image domain, instance discrimination learning is now proving equally convincing in the graph domain, in particular for the node classification task. However, fewer contributions have tackled the link prediction task, the focus of the current paper, where we propose to adapt existing methods in this context. We first provide a rigorous evaluation of existing self-supervised models in the field of link prediction, demonstrating that the main performance depends on the augmentation process for all instance discrimination methods. We then propose a new structural augmentation based on the community structure that is relevant for link prediction. Finally, we introduce two new models, L-GRACE and L-BGRL, that improve the performance of the existing methods, and we show that they perform on par with both supervised and self-supervised models of the current state of the art.
[ ] TODO Add ArXiV Paper link
date: February 2025 to August 2025

This project aims to create a framework for reward function generation in Gymnasium environments, utilizing and VideoLLM in order to enchance aligment too. Below, you can see how the framework is structured. The paper was accepted on LLAIS workshop at ECAI 2025.
arXiv paper
repository github
date: December 2024 to June 2025

A Plateau Ballancing problem with arduino.

A deep analysis on stack Overflow dataset

A study of how work RNN, LSTM, in the case of finding the good Emotion associated to a sentences

The project aims is to predict link from a graphml files using GNN.

We've compared shallow networks, MLP and CNN on MNIST dataset, with a focus on hyperparameter-tuning thanks to optuna.

this project is an ABA-generator and Roberta implementation to classify arguments from kialo.

this project use Spring, nodejs, vuejs. it's a web mobile game.

Report making the state of the art about deep learning methods and they use health.

In this project we've build a state of the art about the multi-view Clustering.

overview of unsupervised (Kmeans, HCA) and supervised (RandomForest, MLP, Adaboost...) learning.

In this project we have study design pattern, and we've buid a chatbot Eliza.