现代电子技术2025，Vol.48Issue(13)：138-144,7.DOI:10.16652/j.issn.1004-373x.2025.13.020
基于GAIL方法的鱼类个体运动策略恢复方法
Individual fish movement strategy recovering approach based on GAIL
摘要
Abstract
Reinforcement learning in fish behavior strategies faces limitations such as being constrained by predefined rules,reward functions relying on prior knowledge,and an inability to fully capture object behavior strategies.In view of this,a method based on generative adversarial imitation learning(GAIL)is proposed to recover individual movement strategies by fish swarm movement trajectory data.The state and action representations of individual fish are designed,and the decision-making process of fish movement is expressed with a fully connected neural network.Experiments were conducted with one learner and multiple individual teachers who navigate with the Vicsek model.Experimental results demonstrate that the GAIL method can recover individual fish movement strategies effectively,providing an efficient strategy learning approach applicable to the study and simulation of other biological swarm behaviors.In-depth analysis of the swarm behavior reveals the interaction rules of individuals and group dynamics.Therefore,the proposed method offers new insights for the application of artificial intelligence in biological behavior research.关键词
生成对抗模仿学习/鱼类集群行为/运动策略恢复/人工智能应用/Vicsek模型/全连接神经网络Key words
GAIL/fish school behavior/movement strategy recovery/artificial intelligence application/Vicsek model/fully connected neural network分类
信息技术与安全科学引用本文复制引用
宋婧菡,陈鹏宇,徐俊,岳圣智,闵中原,刘晓阳,林远山..基于GAIL方法的鱼类个体运动策略恢复方法[J].现代电子技术,2025,48(13):138-144,7.基金项目
辽宁省属本科高校基本科研业务费专项资金资助(2024JBZDZ004) （2024JBZDZ004）
2023中央财政对辽宁渔业补助项目 （）
辽宁省重点研发计划(2023JH26/10200015) （2023JH26/10200015）
辽宁省自然基金资助计划(2020-KF-12-09) （2020-KF-12-09）
辽宁省教育厅基本科研项目(LJKZ0730,QL202016) （LJKZ0730,QL202016）
设施渔业教育部重点实验室开放课题(202219) （202219）
辽宁省应用基础计划项目(2022JH2/101300187) （2022JH2/101300187）
