Multiagent sequential decision making under uncertainty (MSDM) is the problem multiple agents face when they aim to optimise their decisions over a finite or infinite number of time steps in a stochastic environment. It becomes increasingly complex when there is limited communication among the agents and the agents would act subjectively in a real-time interaction. The research problem is one of the core challenging issues in the agents planning community where researchers have developed a significant amount of research in the past decades. For example, a number of multiagent decision models, e.g. Dec-POMDP, MMDP and I-POMDP, have been well developed while a variety of different algorithms have been crafted to solve the models with the consideration of tractable computational complexity. However, the research challenges still exist particularly in an open AI world where unknown unknowns further complicate the agents’ interactions in both offline agent model and online plan execution.
The purpose of this workshop is to revisit the previous research achievements and discuss potential solutions to the new challenges in the MSDM research. Meanwhile, it is noticed that data-driven AI with powerful deep learning have dramatically changed the AI research horizon, which also offers great opportunities to re-shape the MSDM research. Hence, this workshop aims to bring together researchers in the fields of agents planning, reasoning and learning, as well as general machine learning and data science. We expect to discuss new MSDM solutions and foster research collaboration so as to demonstrate the MSDM research in a set of impactful real-world applications, e.g. health, space, security and so on.
This workshop will cover the following main topics:
- Agent planning, reasoning and learning
- Modelling other agents and self-modelling
- New multiagent decision and planning models
- Data-driven agent systems
- Interdisciplinary agent research
- Markov games- MSDM applications
- MSDM Benchmarks and evaluation methodologies
- Future MSDM research and new challenges
We welcome Technical Papers (up to 8 pages) and Short Position Papers (2-4 pages). Papers should be written in English (with clear authors and affiliations), be submitted as a PDF document, and conform to the formatting guidelines of AAMAS 2023:https://aamas2023.soton.ac.uk/wp-content/uploads/sites/443/2022/06/AAMAS-2023-Formatting-Instructions.zip.
The paper Submission Link is:https://easychair.org/conferences/?conf=msdm23.
The workshop articles will be invited for a submission in a journal special issue (TBD).
| February 10, 2023(AoE) | Submission Deadline |
| March 20, 2023 | Decision Notification |
| March 31, 2023(AoE) | Camera Ready |
| May 30, 2023 | MSDM Workshop Dates |
The MSDM-23 workshop will take place jointly with AAMAS 2023 on May 30 2023 in the London ExCeL conference centre, London, United Kingdom.
| Time: 30 May, 2023 | Room: South Gallery 9+10 |
| 9:00-9:15 | Welcome Talk |
| 9:15-10:00 | Invited Talk by Matthijs Spaan (Delft University of Technology, The Netherlands.) |
| 10:00–10:45 | Coffee break with AAMAS |
| 10:45–12:30 | Paper Presentation (25mins) X 5 |
| • Yue Guan, Daigo Shishika, Jason Marden, Panagiotis Tsiotras and Vijay Kumar, Dynamic Adversarial Resource Allocation: A Complete Characterization | |
| • Srijoni Majumdar, Chuhao Qin and Evangelos Pournaras, Discrete-choice Multi-agent Optimization: Decentralized Hard Constraint Satisfaction for Smart Cities | |
| • Adam Michalski, Filippos Christianos and Stefano Albrecht, SMAClite: A Lightweight Environment for Multi-Agent Reinforcement Learning | |
| • Mingyang Sun, Yaqing Hou, Jie Kang, Haiyin Piao, Yifeng Zeng, Hongwei Ge and Qiang Zhang, Improving Cooperative Multi-Agent Exploration via Surprise Minimization and Social Influence Maximization | |
| • James Chao, Wiktor Piotrowski, Mitch Manzanares and Douglas Lange, Novelty Accommodating Multi-Agent Planning in High Fidelity Simulated Open World | |
| 12:30–14:00 | Lunch Break |
| 14:00–15:45 | Paper Presentation (25mins) X 4 |
| • Hannah Tawashy and Prashant Doshi, Recurrent Sum-Product-Max Networks for Multi-Agent Decision Making: A Perspective | |
| • Aurélien Delage, Olivier Buffet, Jilles Steeve Dibangoye and Abdallah Saffidine, Heuristic Search Value Iteration can solve zero-sum Partially Observable Stochastic Games | |
| • Elliot Fosong, Muhammad Arrasy Rahman, Ignacio Carlucho and Stefano Albrecht, Learning Complex Teamwork Tasks using a Sub-task Curriculum | |
| • Sarit Adhikari and Piotr Gmytrasiewicz, Bayesian Learning for literal POMDP in Communicative IPOMDPs | |
| 15:45-16:30 | Coffee break with AAMAS |
| 16:30–17:30 | Panel Discussion |
| Topic: How can MSDM research help improve ChatGPT type agents? | |
| 17:45 | Workshop Close |
| [#1] | Sarit Adhikari and Piotr Gmytrasiewicz, Bayesian Learning for literal POMDP in Communicative IPOMDPs [Link to Paper] |
| [#2] | Yue Guan, Daigo Shishika, Jason Marden, Panagiotis Tsiotras and Vijay Kumar, Dynamic Adversarial Resource Allocation: A Complete Characterization [Link to Paper] |
| [#3] | Aurélien Delage, Olivier Buffet, Jilles Steeve Dibangoye and Abdallah Saffidine, Heuristic Search Value Iteration can solve zero-sum Partially Observable Stochastic Games [Link to Paper] |
| [#4] | James Chao, Wiktor Piotrowski, Mitch Manzanares and Douglas Lange, Novelty Accommodating Multi-Agent Planning in High Fidelity Simulated Open World [Link to Paper] |
| [#5] | Adam Michalski, Filippos Christianos and Stefano Albrecht, SMAClite: A Lightweight Environment for Multi-Agent Reinforcement Learning [Link to Paper] |
| [#6] | Mingyang Sun, Yaqing Hou, Jie Kang, Haiyin Piao, Yifeng Zeng, Hongwei Ge and Qiang Zhang, Improving Cooperative Multi-Agent Exploration via Surprise Minimization and Social Influence Maximization [Link to Paper] |
| [#7] | Elliot Fosong, Muhammad Arrasy Rahman, Ignacio Carlucho and Stefano Albrecht, Learning Complex Teamwork Tasks using a Sub-task Curriculum [Link to Paper] |
| [#8] | Srijoni Majumdar, Chuhao Qin and Evangelos Pournaras, Discrete-choice Multi-agent Optimization: Decentralized Hard Constraint Satisfaction for Smart Cities [Link to Paper] |
| [#9] | Hannah Tawashy and Prashant Doshi, Recurrent Sum-Product-Max Networks for Multi-Agent Decision Making: A Perspective [Link to Paper] |
All questions about submissions should be emailed to Yifeng Zeng (Professor), Northumbria University, UKyifeng.zeng@northumbria.uk.