Unverified50% confidenceFactExact time
鲁棒PCA通过将数据矩阵分解为低秩矩阵加稀疏矩阵（L+S分解）将异常值隔离到稀疏部分
1
Sources
50%
Confidence
Long-term
Relevance
8/23/2026
First Seen
Sources
Entropic Scree：用信息熵替代方差重构PCA降维方法
Related Entities
Related Claims
UnverifiedPrincipal Component Analysis (PCA) involves both eigenvalue decomposition from linear algebra and variance maximization from statistics.64% similarUnverifiedKV cache reuses the Key and Value matrices of historical tokens during the inference phase, avoiding repeated attention computation on already-generated sequences.59% similarUnverifiedPCA can only capture linear correlations and cannot handle nonlinear structures in data manifolds.56% similarUnverified在特征标准化、归一化或PCA降维时使用整个数据集（包括测试集）的统计量而非仅用训练集，会造成特征构造泄露55% similarUnverifiedRRF (Reciprocal Rank Fusion) uses only the ranking position information from each retrieval path (formula RRF(d)=Σ1/(k+rank_i(d)), where k is typically set to 60), is insensitive to score distribution, and bypasses the cross-modal score alignment problem54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/789969API
curl https://kongchang.com/api/v1/knowledge/claims/789969MCP
get_claim(id=789969)