Anas Barakat

Anas Barakat

Research Fellow

Singapore University of Technology and Design

I am currently a research fellow at Singapore University of Technology and Design working with Georgios Piliouras and Antonios Varvitsiotis.

My research develops theory and algorithms for learning in strategic, structured, and dynamic environments. I study how autonomous agents learn to act in multi-agent settings where structure, feedback loops and non-standard or non-stationary objectives shape long-term behavior. My work integrates tools from reinforcement learning, game theory, online learning, stochastic optimization, and dynamical systems to build principled and robust multi-agent learning systems. See below for a more detailed overview of my research.

Previously, I was a postdoctoral fellow at ETH Zurich working with Niao He in the department of computer science. I obtained my PhD in applied mathematics and computer science from Institut Polytechnique de Paris at Télécom Paris under the supervision of Pascal Bianchi and Walid Hachem. I received my engineering Master's degree in applied mathematics and computer science from Télécom Paris and a Master's degree in data science from Université Paris Saclay.

Here is my CV for more information.

Interests

  • Multi-Agent Learning
  • Reinforcement Learning
  • Stochastic Optimization

Education

  • PhD in Applied Mathematics and CS, 2021

    Institut Polytechnique de Paris (Télécom Paris)

  • MSc in Data Science, 2018

    Université Paris Saclay

  • MSc in Applied Mathematics and CS, 2018

    Télécom Paris

Research

Strategic Learning in Structured and Dynamic Multi-Agent Environments
with General Utility and Behaviorally Aligned Objectives

My research develops theory and algorithms for learning in strategic, structured and dynamic environments. I study how autonomous agents learn to act in multi-agent settings where feedback loops, structure, non-standard and non-stationary objectives shape long-term behavior. My work integrates tools from reinforcement learning, game theory, online learning, stochastic optimization, and dynamical systems to build principled and robust multi-agent learning systems.

Below is an overview of my main research directions, with corresponding publications and preprints.

- Multi-Agent Learning in Strategic, Structured and Dynamic Environments

I study learning in multi-agent environments where agents interact strategically—whether cooperatively or competitively. My work investigates decentralized learning dynamics and convergence to equilibria in structured settings. These include structured games such as (constrained) Markov potential games for multi-agent RL and zero-sum linear quadratic games, structured strategy spaces defined via generalized simplices in symmetric cones, and multi-agent linear dynamical systems with arbitrary disturbances and non-stationary time-varying objectives.

- Beyond Expected Returns: General Utility and Behaviorally Aligned RL

I develop theory and algorithms for policy optimization beyond standard additive cumulative rewards, including non-linear and reward-free objectives. Drawing on insights from psychology and behavioral economics, I introduced a policy gradient framework based on cumulative prospect theory to model cognitive and psychological biases in sequential decision-making and capture the diversity of human attitudes toward risk. This work contributes to the design of human-centered RL systems that better align with complex preferences. Applications include personalized healthcare, where psychological factors influence clinical decisions, energy, finance, traffic management as well as domains such as legal and ethical decision-making, cybersecurity, and human-robot interaction.

- A Dynamical Systems and Stochastic Approximation View of Adaptive Optimization

During my PhD, I studied the theoretical properties of adaptive gradient methods (e.g., Adam, Adagrad, RMSProp) using dynamical systems and stochastic approximation techniques, providing unified insights into their convergence and stability.

See below my list of publications organized along these complementary research directions.

Research

Multi-Agent Learning in Strategic, Structured and Dynamic Environments

Anas Barakat, John Lazarsfeld, Georgios Piliouras, Antonios Varvitsiotis (2025). Multi-Agent Online Control with Adversarial Disturbances. Under review.

Arxiv

Anas Barakat, Wayne Lin, John Lazarsfeld, Antonios Varvitsiotis (2025). Optimistic Online Learning in Symmetric Cone Games. Preprint.

Arxiv

Jiduan Wu, Anas Barakat, Ilyas Fatkhullin, Niao He (2025). Learning Zero-Sum Linear Quadratic Games with Improved Sample Complexity and Last Iterate Convergence. To appear in SIAM Journal on Control and Optimization.

Arxiv

Pragnya Alatur, Anas Barakat*, Niao He (2024). Independent Policy Mirror Descent for Markov Potential Games: Scaling to Large Number of Players. IEEE CDC 2024. *corresponding author.

Slides Proceedings Arxiv

Philip Jordan, Anas Barakat, Niao He (2023). Independent Learning in Constrained Markov Potential Games. AISTATS 2024.

Poster Slides Proceedings Arxiv

Jiduan Wu, Anas Barakat, Ilyas Fatkhullin, Niao He (2023). Learning Zero-Sum Linear Quadratic Games with Improved Sample Complexity. IEEE CDC 2023.

Proceedings Arxiv

Research

Expected Returns and Beyond: General Utility and Behaviorally Aligned RL

Olivier Lepel, Anas Barakat (2025). A Prospect-Theoretic Policy Gradient Algorithm for Behavioral Alignment in Reinforcement Learning. Under review.

Slides Arxiv

Anas Barakat, Souradip Chakraborty, Peihong Yu, Pratap Tokekar, Amrit Singh Bedi (2024). Towards Scalable General Utility Reinforcement Learning: Occupancy Approximation, Sample Complexity and Global Optimality. Under review.

Arxiv

Kimon Protopapas, Anas Barakat (2024). Policy Mirror Descent with Lookahead. NeurIPS 2024.

Proceedings Arxiv

Anas Barakat, Ilyas Fatkhullin, Niao He (2023). Reinforcement Learning with General Utilities: Simpler Variance Reduction and Large State-Action Space. ICML 2023.

Poster Slides Proceedings Arxiv

Ilyas Fatkhullin, Anas Barakat, Anastasia Kireeva, Niao He (2023). Stochastic Policy Gradient Methods: Improved Sample Complexity for Fisher-non-degenerate Policies. ICML 2023.

Poster Proceedings Arxiv

Anas Barakat, Pascal Bianchi, Julien Lehmann (2022). Analysis of a Target-Based Actor-Critic Algorithm with Linear Function Approximation. AISTATS 2022.

Poster Proceedings Arxiv

Research

A Dynamical Systems and Stochastic Approximation View of Adaptive Optimization

Anas Barakat, Pascal Bianchi, Walid Hachem, Sholom Schechtman (2021). Stochastic optimization with momentum: convergence, fluctuations, and traps avoidance. Electronic Journal of Statistics 15 (2), 3892-3947.

DOI Arxiv Journal

Anas Barakat, Pascal Bianchi (2020). Convergence and Dynamical Behavior of the ADAM Algorithm for Non-Convex Stochastic Optimization. SIAM Journal on Optimization, 31 (1), 244-274.

Poster Slides Video DOI Arxiv Journal

Anas Barakat, Pascal Bianchi (2019). Convergence Rates of a Momentum Algorithm with Bounded Adaptive Step Size for Non-Convex Optimization. ACML 2020.

Poster Slides Proceedings Arxiv

Talks

Invited talk - Learning Theory and Applications Workshop, Nanyang Technological University
Apr 29, 2025 4:00 PM College of Computing and Data Science, NTU, Singapore
Invited talk - Finance and RL Talks
Apr 21, 2025 10:00 PM Online
Invited talk - 4th Symposium on Machine Learning and Dynamical Systems
Jul 8, 2024 2:45 PM Fields Institute, Toronto, Canada
Talk - SUTD Group Meeting
May 3, 2024 10:00 AM Online
Invited talk - ICCOPT 2022
Jul 28, 2022 3:00 PM Lehigh University, Bethlehem, USA
See all talks

Reviewing

  • Conferences: NeurIPS, ICML, ICLR, COLT, AISTATS, IEEE CDC.

  • Journals: Journal of Machine Learning Research (JMLR), Transactions on Machine Learning Research (TMLR), Mathematical Programming, SIAM Journal on Optimization (SIOPT), Journal of Optimization Theory and Applications (JOTA), Systems & Control Letters, Mathematics of Control, Signals, and Systems (MCSS), Stochastic Systems, IEEE Transactions on Automatic Control, IEEE Control Systems Letters.

Teaching

ETH Zurich (2022-2024):

Télécom Paris (2018-2021): Teaching Assistant

Contact