Unverified50% confidenceFactExact time
研究团队证明学习分布与真实分布之间的总变差距离可分解为初始化误差、Score误差和求解器离散化误差三项
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
Related Claims
UnverifiedBehavioral cloning simplifies the problem to supervised learning, but suffers from distribution shift, where differences between the state distributions at training time and execution time lead to error accumulation.77% similarUnverified行为克隆（BC）存在复合误差与分布偏移问题，智能体偏离训练分布后预测误差会指数级累积77% similarUnverified数值实验表明三项误差中score误差项占据主导，整体TV误差与score误差呈线性关系76% similarVerifiedBehavioral cloning suffers from the distribution shift problem, where the model tends to make errors and accumulate mistakes on states outside the training distribution, a phenomenon referred to as compounding error in the robotics control field.74% similarUnverified长链推理中每步的微小误差会被累积放大，学术界称为复合错误（compounding errors）问题73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/514882API
curl https://kongchang.com/api/v1/knowledge/claims/514882MCP
get_claim(id=514882)