Youssef Mroueh

Youssef Mroueh

Research Staff Member

IBM Research and MIT-IBM Watson AI Lab

Biography

I am a Principal Research Scientist in IBM Research and a principal investigator in the MIT-IBM Watson AI lab. I joined IBM in 2015, after finishing my PhD in computer science in February 2015 from MIT, CSAIL, where I was advised by Professor Tomaso Poggio. In 2011, I obtained my engineering diploma from Ecole Polytechnique Paris France, and a master of science in Applied Maths from Ecole des Mines de Paris.

Interests

  • Deep Learning and Generative Modeling
  • Optimal Transport
  • Statistical learning theory
  • Multimodal Learning, vision and language
  • Image captioning As an Assistive Technology
  • Generative models for accelerated scientific discoveries

Recent Publications

Unsupervised Hierarchy Matching with Optimal Transport over Hyperbolic Spaces

This paper focuses on the problem of unsupervised alignment of hierarchical data such as ontologies or lexical databases. This problem …

ACM Books - Article/Chapter View

PDF

Unbalanced Sobolev Descent

We introduce Unbalanced Sobolev Descent (USD), a particle descent algorithm for transporting a high dimensional source distribution to …
PDF Code
See all publications

Recent Posts

Information Theoretic Limits of Alignment in Large Language Models

1. The Alignment Problem Reinforcement Learning with Human Feedback (RLHF): Constrained Policy Optimization The alignment problem aims to align large language models (LLMs) with desired behaviors using constrained policy optimization. The objective is to find a policy that maximizes a reward $r$ while staying close to a reference policy $ \pi_{\text{ref}}$.
Last updated on Nov 1, 2024 3 min read