Seth Karten

Seth Karten

I am a PhD Candidate in Computer Science at Princeton University advised by Chi Jin. My current research interests involve studying multi-agent interaction and open-ended learning at scale to enable autonomous agents. I am interested in using foundation models (such as large language models) and reinforcement learning for decentralized decision-making that can scale individual preferences to group behavior.

Previously, I received an MS in Robotics from Carnegie Mellon University, advised by Katia Sycara, where I studied emergent communication and decision-making in multi-agent teams. I obtained my undergraduate degree from Rutgers University, New Brunswick in Computer Science and Mathematics, where I received the C. Greg Hagerty Artificial Intelligence and Computer Science Award. I studied learning hierarchical control primitives under the supervision of Kostas Bekris. I previously spent some time as an Applied Scientist at Amazon studying multi-agent pathfinding (MAPF). I am a recipient of the NSF Graduate Research Fellowship and Francis Robbins Upton Fellowship.

I am always excited to collaborate with others. If you are interested in open-ended learning or decentralized multi-agent teams in the era of foundation models, please do not hesitate to reach out.

✉️ Email  /  📄 CV  /  📚 Google Scholar  /  X  /  BlueSky

News

December 14, 2024: I presented a poster on PokéChamp: An Expert-level Minimax Language Agent for Competitive Pokémon at the NeurIPS 2024 Language Gamification Workshop.

June 19, 2024: I presented a poster on The LLM Economist: Optimizing Policy in Multiagent Generative Simulations at the Cooperative AI Summer School.

March 29, 2023: I am excited to announce that I have been awarded the National Science Foundation Graduate Research Fellowship (NSF GRFP)!

Selected Publications

Language Agents

PokéChamp: An Expert-level Minimax Language Agent for Competitive Pokémon
Seth Karten, Andy Nguyen, Chi Jin
NeurIPS Language Gamification Workshop, 2024
PDF Code Website

Competitive Games

FightLadder: A Benchmark for Competitive Multi-Agent Reinforcement Learning
Wenzhe Li, Zihan Ding, Seth Karten, Chi Jin
ICML, 2024

Emergent Communication

On the Role of Emergent Communication for Social Learning in Multi-Agent Reinforcement Learning
Seth Karten, Siva Kailas, Huao Li, Katia Sycara
AAMAS, 2023
Previous version appeared in IROS 2022 Workshop on Decision Making in Multi-Agent Systems.

Towards True Lossless Sparse Communication in Multi-Agent Systems
Seth Karten, Mycal Tucker, Siva Kailas, Katia Sycara
ICRA, 2023
Previous version appeared in NeurIPS 2022 Workshop on Deep Reinforcement Learning.

Interpretable Learned Emergent Communication for Human-Agent Teams
Seth Karten, Mycal Tucker, Huao Li, Siva Kailas, Michael Lewis, Katia Sycara
IEEE Transactions on Cognitive and Developmental Systems, 2023
Previous version appeared in IROS 2022 Workshop on Human Theory of Machines and Machine Theory of Mind for Human-Agent Teams.

Autonomous Vehicles

Improving Kinodynamic Planners for Vehicular Navigation with Learned Goal-Reaching Controllers
Aravind Sivaramakrishnan, Edgar Granados, Seth Karten, Troy McMahon, Kostas E. Bekris
IROS, 2021
Previous version appeared in ICRA 2021 Machine Learning for Motion Planning Workshop.