Loughborough University
Browse
128_sharing_lifelong_reinforcement.pdf (5.73 MB)

Sharing lifelong reinforcement learning knowledge via modulating masks

Download (5.73 MB)
conference contribution
posted on 2023-07-05, 13:58 authored by Saptarshi NathSaptarshi Nath, Christos PeridisChristos Peridis, Eseoghene Ben-Iwhiwhu, Xinran Liu, Shirin DoraShirin Dora, Cong Liu, Soheil Kolouri, Andrea SoltoggioAndrea Soltoggio

Lifelong learning agents aim to learn multiple tasks sequentially over a lifetime. This involves the ability to exploit previous knowledge when learning new tasks and to avoid forgetting. Recently, modulating masks, a specific type of parameter isolation approach, have shown promise in both supervised and reinforcement learning. While lifelong learning algorithms have been investigated mainly within a single-agent approach, a question remains on how multiple agents can share lifelong learning knowledge with each other. We show that the parameter isolation mechanism used by modulating masks is particularly suitable for exchanging knowledge among agents in a distributed and decentralized system of lifelong learners. The key idea is that isolating specific task knowledge to specific masks allows agents to transfer only specific knowledge on-demand, resulting in a robust and effective collective of agents. We assume fully distributed and asynchronous scenarios with dynamic agent numbers and connectivity. An on-demand communication protocol ensures agents query their peers for specific masks to be transferred and integrated into their policies when facing each task. Experiments indicate that on-demand mask communication is an effective way to implement distributed and decentralized lifelong reinforcement learning, and provides a lifelong learning benefit with respect to distributed RL baselines such as DD-PPO, IMPALA, and PPO+EWC. The system is particularly robust to connection drops and demonstrates rapid learning due to knowledge exchange.

Funding

Defense Advanced Research Projects Agency (DARPA) under Contract No. HR00112190132 (Shared Experience Lifelong Learning)

History

School

  • Science

Department

  • Computer Science

Published in

Proceedings of the 2nd Conference on Lifelong Learning Agents (CoLLAs 2023)

Pages

936 - 960

Source

Second Conference on Lifelong Learning Agents (CoLLAs 2023)

Publisher

ML Research Press

Version

  • VoR (Version of Record)

Rights holder

© The Authors

Publisher statement

This paper appears here with the permission of the conference program committee.

Acceptance date

2023-05-15

Copyright date

2023

ISSN

2640-3498

Book series

Proceedings of Machine Learning Research; volume 232

Language

  • en

Editor(s)

Sarath Chandar; Razvan Pascanu; Hanie Sedghi; Doina Precup

Location

Montreal, Canada

Event dates

22nd August 2023 - 25th August 2023

Depositor

Dr Andrea Soltoggio. Deposit date: 1 July 2023

Usage metrics

    Loughborough Publications

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC