Muhammad Arrasy Rahman

Postdoctoral Research Fellow at the Learning Agents Research Group.

prof_pic.jpeg

Gates-Dell Complex Room 3.418

2317 Speedway

The University of Texas at Austin

Austin, Texas 78712

I am a Postdoctoral Research Fellow at Professor Peter Stone’s Learning Agents Research Group (LARG) at UT Austin. Prior to my current position, I obtained my masters and doctorate degree from the University of Edinburgh, where I was a member of the Autonomous Agents Research group headed by Dr. Stefano Albrecht. Originally, I was born in Indonesia and lived there until I obtained my undergraduate degree from Universitas Indonesia.

My research interests broadly encompass game theory, reinforcement learning, transfer learning and graph neural networks. I am specifically interested in their application in the ad hoc teamwork (AHT) problem, which explores techniques to create autonomous agents that can quickly adapt their policies to optimally interact with previously unseen teammate policies and team configurations. Through my research, my long-term goal is to build intelligent agents that can collaborate on-the-fly with humans to help them solve various collaborative decision-making problems encountered in their daily lives.

At the moment, I am exploring methods to generate a diverse population of teammate policies that can be utilized to train robust agents capable of optimally collaborating against different teammate policies. My belief is the trained agent’s robustness can be maximized if we design a population of teammate policies requiring the trained agent to play distinct best-response policies. If you are also interested in this area of research, please do not hesitate to contact me.

News

Mar 10, 2024 :newspaper_roll: Our latest work on instilling teamwork in agents via subtask curriculum has now been accepted at AAMAS 2024: Learning Complex Teamwork Tasks Using a Given Sub-task Decomposition
Dec 10, 2023 :newspaper_roll: My latest work on generating diverse teammate policies has now been accepted at AAAI 2024: Minimum Coverage Sets for Training Robust Ad Hoc Teamwork Agents
Oct 23, 2023 Together with my fellow AHT researchers, we are starting a new seminar series to disseminate current works in AHT, zero-shot coordination, agent modelling, and human-robot interaction. Check our website to find more information of upcoming meetings and presenters.
Oct 20, 2023 I’m excited to announce the start of a new partnership with Lockheed Martin Corporation. In this collaboration, we will explore methods to generate teammate policies to create more robust AHT agents.
Sep 25, 2023 We are going to organize a workshop at AAAI-24. Check out the details for our upcoming workshop here.

Selected Publications

2024

  1. Minimum Coverage Sets for Training Robust Ad Hoc Teamwork Agents
    Muhammad Rahman, Jiaxun Cui, and Peter Stone
    Proceedings of the AAAI Conference on Artificial Intelligence, Mar 2024

2023

  1. BRDiv.png
    Generating Teammates for Training Robust Ad Hoc Teamwork Agents via Best-Response Diversity
    Arrasy Rahman, Elliot Fosong, Ignacio Carlucho, and 1 more author
    Transactions on Machine Learning Research, Mar 2023
  2. JMLR
    A General Learning Framework for Open Ad Hoc Teamwork Using Graph-based Policy Learning
    Arrasy Rahman, Ignacio Carlucho, Niklas Höpner, and 1 more author
    Journal of Machine Learning Research, Mar 2023
  3. arXiv
    Minimum Coverage Sets for Training Robust Ad Hoc Teamwork Agents
    Arrasy Rahman, Jiaxun Cui, and Peter Stone
    Mar 2023

2022

  1. EUMAS
    A survey of ad hoc teamwork research
    Reuth Mirsky, Ignacio Carlucho, Arrasy Rahman, and 5 more authors
    In European Conference on Multi-Agent Systems, Mar 2022
  2. arXiv
    A General Learning Framework for Open Ad Hoc Teamwork Using Graph-based Policy Learning
    Arrasy Rahman, Ignacio Carlucho, Niklas Höpner, and 1 more author
    arXiv preprint arXiv:2210.05448, Mar 2022

2021

  1. GPL.png
    Towards open ad hoc teamwork using graph-based policy learning
    Muhammad A Rahman, Niklas Hopner, Filippos Christianos, and 1 more author
    In International Conference on Machine Learning, Mar 2021