Verified75% confidenceFactExact time
The risk-coverage curve plots the proportion of cases automatically handled by the model on the x-axis and the error risk at that coverage level on the y-axis; a smaller area under the curve indicates more reliable judgments in high-confidence intervals.
3
Sources
75%
Confidence
Long-term
Relevance
10/5/2026
First Seen
Sources
Related Entities
Related Claims
Unverified未校准的模型若直接套用固定阈值做自动决策，会系统性地高估或低估风险，因此上线前应绘制可靠性图并评估ECE等指标。73% similarUnverifiedAdversarial examples often lie at the edge of or outside ray coverage, while mislabeled samples may be covered within the expanded regions of training samples, causing the model to "confidently" give wrong predictions.73% similarUnverifiedWhen samples are extremely imbalanced, PR curves more accurately reflect model performance on the minority class, and when evaluating fraud detection capability, PR curves and AUPRC metrics are relied upon more heavily.71% similarUnverifiedThe model output may not be a simple "yes/no" determination but a risk score, which is applied with different thresholds in different scenarios (more lenient for real-time developer reminders, stricter for security team audits).71% similarVerifiedSome models carry the risk of over-optimization toward test sets, causing leaderboard scores to deviate from real-world performance (benchmark saturation).71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/975452API
curl https://kongchang.com/api/v1/knowledge/claims/975452MCP
get_claim(id=975452)