Skip to main navigation Skip to search Skip to main content

A Model Compression Method for Deep Reinforcement Learning

  • Xin Deng
  • , Lei Chen*
  • *Corresponding author for this work
  • Beijing Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Deep reinforcement learning (DRL) achieves remarkable success in complex tasks, but its large model size makes it a major challenge to deploy it on resource-constrained platforms while maintaining high performance and achieving high compression ratio. In this paper, we propose a DRL model compression method that combines group regularization pruning and random sketches. We first verify the effectiveness in a classic DRL environment and then demonstrate its application value in a multi-agent confrontation environment. Experimental results show that our method achieves a high compression ratio and superior policy performance over baselines.

Original languageEnglish
Title of host publicationProceedings of 2025 9th Chinese Conference on Swarm Intelligence and Cooperative Control - Swarm Optimization Technologies
EditorsYongzhao Hua, Yishi Liu, Rui Yan
PublisherSpringer Science and Business Media Deutschland GmbH
Pages174-185
Number of pages12
ISBN (Print)9789819583287
DOIs
Publication statusPublished - 2026
Event9th Chinese Conference on Swarm Intelligence and Cooperative Control, CCSICC 2025 - Shanghai, China
Duration: 31 Oct 20253 Nov 2025

Publication series

NameLecture Notes in Electrical Engineering
Volume1606 LNEE
ISSN (Print)1876-1100
ISSN (Electronic)1876-1119

Conference

Conference9th Chinese Conference on Swarm Intelligence and Cooperative Control, CCSICC 2025
Country/TerritoryChina
CityShanghai
Period31/10/253/11/25

Keywords

  • Deep reinforcement learning
  • Group regularization pruning
  • Multi-agent systems
  • Random sketches

Fingerprint

Dive into the research topics of 'A Model Compression Method for Deep Reinforcement Learning'. Together they form a unique fingerprint.

Cite this