When Does Predictive Inverse Dynamics Outperform Behavior Cloning?
何时预测性逆动力学优于行为克隆?
机构 * University of Cambridge(剑桥大学) ; Universitygrow
AI总结 本文通过理论分析解释了预测性逆动力学模型(PIDM)为何在行为克隆(BC)失败时表现更优,归因于偏差-方差权衡,并实验验证了PIDM在样本效率上的显著优势。
Comments To be published in proceedings of the International Conference on Machine Learning (ICML), 2026