Unverified50% confidenceFactExact time
RETD在两状态反例和一个Baird点上均实现了经过认证的负李雅普诺夫指数，而Baird反例中ETD的正指数符号仍停留在数值层面
1
Sources
50%
Confidence
Long-term
Relevance
9/18/2026
First Seen
Sources
Related Entities
Related Claims
Partially VerifiedAI领域常采用0到1的连续值打分，也可用0和1两个离散值表示传统的不通过与满分61% similarUnverified违反同方差假设后，OLS 参数估计仍无偏，但不再具有最小方差，且标准误差估计会出错58% similarUnverified逻辑回归的损失函数（交叉熵）本质上就是负对数似然57% similarUnverified岭回归通过在损失函数中添加L2正则化项将回归系数向零收缩但不置零，而LASSO（L1正则化）会产生稀疏解57% similarUnverified困惑度等于测试集交叉熵损失的指数：PPL = exp(loss)，loss=1.77对应PPL≈5.8657% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/931064API
curl https://kongchang.com/api/v1/knowledge/claims/931064MCP
get_claim(id=931064)