Unverified50% confidenceFactExact time
Platt Scaling通过Sigmoid函数P(y=1|x)=1/(1+exp(Af(x)+B))映射概率，参数A和B通过最大似然估计在留出集上拟合
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
Related Claims
Unverifiedsklearn的SVC使用John Platt于1999年提出的Platt Scaling方法，在SVM训练完成后额外拟合一个Sigmoid函数将决策值映射到[0,1]区间来输出概率62% similarUnverified如果流式响应计费实现有误，将每个 chunk 携带的累计 Token 数当作增量叠加，会产生等差数列求和 n(n+1)/2 的放大效应60% similarUnverified差分启发式的查询阶段计算 h(n, t) = 所有地标L中 |dist(n, L) - dist(t, L)| 的最大值，且该估计永远不会超过真实距离（可采纳性成立）59% similarUnverifiedProduct quantization converts distance computation from per-dimension floating-point operations into integer table-lookup accumulation using precomputed codebook distances, thereby improving batch retrieval throughput58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/558797API
curl https://kongchang.com/api/v1/knowledge/claims/558797MCP
get_claim(id=558797)