PBT-GAIL: An Imitation Learning Framework in Swarm Systems

You are here: Homepage > Search

PBT-GAIL: An Imitation Learning Framework in Swarm Systems

Liu, Shuo / Peng, Xingguang / Wang, Tonghao

The swarm system is a kind of multi-agent system where amazing “emergence” phenomenon can be observed. In recent years, biologists have made great effort to model the self-organizing rules of collective behaviors. By analyzing the data collected from biological swarm systems, a plenty of models have been built to reproduce and explain the collective behaviors. Data-driven methods are also promising ways to model collective behaviors. Imitation learning (IL), as a data-driven method, can learn the potential model from the expert data. Further, the Generative Adversarial Imitation Learning (GAIL) is a popular imitation learning method and has been used in many fields as auto-driving, machine translation and image generation. In this paper, we introduce a Population-based Training (PBT) method to GAIL to enhance the performance of GAIL. Moreover, we use this PBT-GAIL framework to model the self-organizing rules according to the simulation data of the icsek model. Simulation results show that the proposed PBT-GAL framework is effective when imitating the icsek model. Besides, the PBT-GAIL framework has better convergence tendency and alignment than that of GAIL.

Access

Check access

Check availability in my library

Order at Subito €

Export, share and cite

Document information

Title :

PBT-GAIL: An Imitation Learning Framework in Swarm Systems

Additional title:

Lect. Notes Electrical Eng.

Contributors:

Wu, Meiping (editor) / Niu, Yifeng (editor) / Gu, Mancang (editor) / Cheng, Jin (editor) / Liu, Shuo (author) / Peng, Xingguang (author) / Wang, Tonghao (author)

Conference:

International Conference on Autonomous Unmanned Systems ; 2021 ; Changsha, China September 24, 2021 - September 26, 2021

Published in:

Proceedings of 2021 International Conference on Autonomous Unmanned Systems (ICAUS 2021) ; Chapter : 186 ; 1884-1894

Lecture Notes in Electrical Engineering ; 861

Publication date :

2022-03-18

Size :

11 pages

ISBN :

978-981-16-9492-9 , 978-981-16-9491-2

ISSN :