Unverified50% confidenceFactExact time
The distribution shift problem in behavior cloning causes error accumulation, with the expected total loss after T steps being O(εT²) rather than the linear O(εT).
1
Sources
50%
Confidence
Long-term
Relevance
9/8/2026
First Seen
Sources
Related Entities
Related Claims
Unverified理论分析表明，在时间步长为T的轨迹上，行为克隆的累积误差以O(T²)的速度增长73% similarUnverifiedIn robust regression, replacing the likelihood function from Gaussian to a t-distribution is equivalent to reducing the weight of outliers in the loss function.71% similarUnverifiedAgent链式调用多步之后，每次LLM输出的细微差异导致误差逐步放大，最终结果偏离预期71% similarUnverified自回归生成的误差累积效应源于模型只能沿单一路径前进，误差以乘积形式放大70% similarVerifiedGradient descent iteratively adjusts weights in the opposite direction of the loss function's gradient, causing the prediction error to converge progressively.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/880293API
curl https://kongchang.com/api/v1/knowledge/claims/880293MCP
get_claim(id=880293)