Diverse Behavior
Diverse behavior research focuses on enabling systems, particularly robots and AI agents, to exhibit a wide range of actions and skills adaptable to various situations. Current research emphasizes developing algorithms and models, such as diffusion models, Bayesian optimization, and contrastive learning methods, to efficiently learn and generate diverse behaviors from limited data, often incorporating constraints and human preferences. This field is significant for advancing robotics, AI, and recommendation systems, enabling more robust, adaptable, and user-friendly technologies. The development of effective metrics for quantifying behavioral diversity is also a key area of ongoing investigation.
Papers
Towards Diverse Behaviors: A Benchmark for Imitation Learning with Human Demonstrations
Xiaogang Jia, Denis Blessing, Xinkai Jiang, Moritz Reuss, Atalay Donat, Rudolf Lioutikov, Gerhard Neumann
Personalized Behavior-Aware Transformer for Multi-Behavior Sequential Recommendation
Jiajie Su, Chaochao Chen, Zibin Lin, Xi Li, Weiming Liu, Xiaolin Zheng
Learning Diverse Skills for Local Navigation under Multi-constraint Optimality
Jin Cheng, Marin Vlastelica, Pavel Kolev, Chenhao Li, Georg Martius
AlignDiff: Aligning Diverse Human Preferences via Behavior-Customisable Diffusion Model
Zibin Dong, Yifu Yuan, Jianye Hao, Fei Ni, Yao Mu, Yan Zheng, Yujing Hu, Tangjie Lv, Changjie Fan, Zhipeng Hu