Automatic Curriculum Learning for Large-Scale Cooperative Multiagent Systems

Zhang, Tianle; Liu, Zhen; Pu, Zhiqiang; Yi, Jianqiang

doi:10.1109/tetci.2022.3209655

Cited by 4 publications

(1 citation statement)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here, we compare with four baselines: (1)the SOTA MARL non-curriculum algorithm HPN-QMIX (Hao et al 2022) (a stronger baseline than our backbone model HPN-VDN) and three MARL curriculum baselines:(2)DYMA (Wang et al 2020), an algorithm that generates task sequences in increments of the number of agents. Two modified versions of PORTAL (3)ERFN-OO and (4)ERFN-MAACL, since the network of OO (da Silva and Costa 2018) and MAACL (Zhang et al 2022) cannot be directly applied in our task scenarios, we only use their task selection criterion and other parts are the same as PORTAL.…”

Section: Comparison With Related Baselines (Rq1)mentioning

confidence: 99%

PORTAL: Automatic Curricula Generation for Multiagent Reinforcement Learning

Wu,

Hao,

Yang

et al. 2024

AAAI

View full text Add to dashboard Cite

Despite many breakthroughs in recent years, it is still hard for MultiAgent Reinforcement Learning (MARL) algorithms to directly solve complex tasks in MultiAgent Systems (MASs) from scratch. In this work, we study how to use Automatic Curriculum Learning (ACL) to reduce the number of environmental interactions required to learn a good policy. In order to solve a difficult task, ACL methods automatically select a sequence of tasks (i.e., curricula). The idea is to obtain maximum learning progress towards the final task by continuously learning on tasks that match the current capabilities of the learners. The key question is how to measure the learning progress of the learner for better curriculum selection. We propose a novel ACL framework, PrOgRessive mulTiagent Automatic curricuLum (PORTAL), for MASs. PORTAL selects curricula according to two critera: 1) How difficult is a task, relative to the learners’ current abilities? 2) How similar is a task, relative to the final task? By learning a shared feature space between tasks, PORTAL is able to characterize different tasks based on the distribution of features and select those that are similar to the final task. Also, the shared feature space can effectively facilitate the policy transfer between curricula. Experimental results show that PORTAL can train agents to master extremely hard cooperative tasks, which can not be achieved with previous state-of-the-art MARL algorithms.

show abstract

Section: Comparison With Related Baselines (Rq1)mentioning

confidence: 99%