Lina Mezghani

I am a final-year PhD student at FAIR Paris and Inria (Thoth team). I am advised by Piotr Bojanowski and Karteek Alahari. My research focuses on making reinforcement learning algorithm rely less on supervision, by exploiting self-supervised and unsupervised methods.

I received an MSc/BSc in Applied Mathematics & Computer Science from École polytechnique (diplôme d'Ingénieur) and an MSc in Machine Learning from École Normale Supérieure Paris-Saclay (Master MVA). During my studies I was a software engineer intern at Pilot Vision, Stockholm, and I visited Cornell Tech University working on computer vision with Prof. Mor Naaman.

Email  /  CV  /  Google Scholar  /  LinkedIn  /  GitHub

profile photo

Learning Goal-Conditioned Policies Offline with Self-Supervised Reward Shaping
Lina Mezghani, Sainbayar Sukhbaatar, Piotr Bojanowski, Alessandro Lazaric, Karteek Alahari
CoRL, 2022
paper / code / project page / bibtex


Walk the Random Walk: Learning to Discover and Reach Goals Without Supervision
Lina Mezghani, Sainbayar Sukhbaatar, Piotr Bojanowski, Karteek Alahari
Agent Learning in Open-Endedness (ALOE) Workshop, ICLR, 2022 (oral)
paper / bibtex


Memory-Augmented Reinforcement Learning for Image-Goal Navigation
Lina Mezghani, Sainbayar Sukhbaatar, Thibaut Lavril, Oleksandr Maksymets, Dhruv Batra, Piotr Bojanowski, Karteek Alahari
IROS, 2022
paper / data / bibtex


Learning to Visually Navigate in Photorealistic Environments Without any Supervision
Lina Mezghani, Sainbayar Sukhbaatar, Arthur Szlam, Armand Joulin, Piotr Bojanowski
arXiv, 2020
paper / bibtex

The code for this page is available here.
Credits to Jon Barron for the template.