Unverified50% confidenceFactExact time
A common feature of non-Bayesian decision rules is that they have discontinuities in parameter space—a tiny change in the estimate may cross a decision boundary, causing the action to jump.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
Unverified浮点运算不满足结合律，不同线程束的执行顺序差异会导致数值结果出现细微偏差78% similarUnverified大模型对指令的敏感度呈非线性分布——主干指令的微小改动可能引发全局行为漂移，而在末尾追加具体的边界条件则相对安全70% similarUnverifiedThe latent jump problem exists for jump-diffusion processes under discrete observation settings, making likelihood-based estimation analytically intractable in most cases70% similarUnverifiedDecision margin or reasoning length can predict which rows will change—the lower the model confidence and the more verbose the reasoning chain, the greater the likelihood of that row flipping68% similarUnverified决策树的分裂准则（信息增益、基尼不纯度、均方误差）本质上是排序统计量的函数，只关心阈值两侧的类别比例而非样本间距离68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1042139API
curl https://kongchang.com/api/v1/knowledge/claims/1042139MCP
get_claim(id=1042139)