Meta-Inverse Reinforcement Learning for Mean Field Games via Probabilistic Context Variables

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Chen, Yang, Lin, Xiao, Yan, Bo, Zhang, Libo, Liu, Jiamou, Tan, Neset Özkan, Witbrock, Michael
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908518940934144
author Chen, Yang
Lin, Xiao
Yan, Bo
Zhang, Libo
Liu, Jiamou
Tan, Neset Özkan
Witbrock, Michael
author_facet Chen, Yang
Lin, Xiao
Yan, Bo
Zhang, Libo
Liu, Jiamou
Tan, Neset Özkan
Witbrock, Michael
contents Designing suitable reward functions for numerous interacting intelligent agents is challenging in real-world applications. Inverse reinforcement learning (IRL) in mean field games (MFGs) offers a practical framework to infer reward functions from expert demonstrations. While promising, the assumption of agent homogeneity limits the capability of existing methods to handle demonstrations with heterogeneous and unknown objectives, which are common in practice. To this end, we propose a deep latent variable MFG model and an associated IRL method. Critically, our method can infer rewards from different yet structurally similar tasks without prior knowledge about underlying contexts or modifying the MFG model itself. Our experiments, conducted on simulated scenarios and a real-world spatial taxi-ride pricing problem, demonstrate the superiority of our approach over state-of-the-art IRL methods in MFGs.
format Preprint
id arxiv_https___arxiv_org_abs_2509_03845
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Meta-Inverse Reinforcement Learning for Mean Field Games via Probabilistic Context Variables
Chen, Yang
Lin, Xiao
Yan, Bo
Zhang, Libo
Liu, Jiamou
Tan, Neset Özkan
Witbrock, Michael
Machine Learning
Artificial Intelligence
Computer Science and Game Theory
Designing suitable reward functions for numerous interacting intelligent agents is challenging in real-world applications. Inverse reinforcement learning (IRL) in mean field games (MFGs) offers a practical framework to infer reward functions from expert demonstrations. While promising, the assumption of agent homogeneity limits the capability of existing methods to handle demonstrations with heterogeneous and unknown objectives, which are common in practice. To this end, we propose a deep latent variable MFG model and an associated IRL method. Critically, our method can infer rewards from different yet structurally similar tasks without prior knowledge about underlying contexts or modifying the MFG model itself. Our experiments, conducted on simulated scenarios and a real-world spatial taxi-ride pricing problem, demonstrate the superiority of our approach over state-of-the-art IRL methods in MFGs.
title Meta-Inverse Reinforcement Learning for Mean Field Games via Probabilistic Context Variables
topic Machine Learning
Artificial Intelligence
Computer Science and Game Theory
url https://arxiv.org/abs/2509.03845