PDJA jointly perturbs perception and decision in multi-agent reinforcement learning. By exploiting state–action synergy, it drives agents into low-reward regions, outperforming single-vector attacks and bypassing existing defenses. (IMAGE)

ELSP

PDJA jointly perturbs perception and decision in multi-agent reinforcement learning. By exploiting state–action synergy, it drives agents into low-reward regions, outperforming single-vector attacks and bypassing existing defenses.

Caption

PDJA jointly perturbs perception and decision in multi-agent reinforcement learning. By exploiting state–action synergy, it drives agents into low-reward regions, outperforming single-vector attacks and bypassing existing defenses.

Credit

Weiqi Guo, Guanjun Liu, Ziyuan Zhou, Department of Computer Science, Tongji University, Shanghai, China

Usage Restrictions

Credit must be given to the creator.

License

CC BY

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.