Unverified50% confidenceFactExact time
When the system matrix is highly non-normal, traditional eigenvalue-based analysis methods severely underestimate transient growth effects, whereas analysis based on the numerical range provides more reliable estimates.
1
Sources
50%
Confidence
Long-term
Relevance
8/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn class-imbalanced scenarios, accuracy alone can be misleading and needs to be combined with a confusion matrix for more fine-grained evaluation.71% similarVerified若用含测试集的全量数据计算均值和标准差，会导致预处理泄漏，使评估指标虚高，无法真实反映模型泛化能力70% similarUnverified结构化、类型强约束的工具响应可以显著降低模型在解析阶段产生误解的概率68% similarUnverified强行取极值样本会引入离群点，迫使SVM构建更复杂的决策边界，高阶poly核在这种数据上收敛更慢67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/822476API
curl https://kongchang.com/api/v1/knowledge/claims/822476MCP
get_claim(id=822476)