Unverified50% confidenceFactExact time
MC-Dropout由Gal和Ghahramani在2016年提出，将其解释为近似贝叶斯推断，通常需要30-100次前向传播才能获得可靠的不确定性估计
1
Sources
50%
Confidence
Long-term
Relevance
8/8/2026
First Seen
Sources
往返一致性：让扩散模型自己预测误差的自监督方法
Related Claims
UnverifiedThe credible interval in Bayesian methods can be directly interpreted as the probability that a parameter falls within the range, providing a more intuitive quantification of uncertainty than the frequentist confidence interval.62% similarUnverifiedThe core challenge of MLOps lies in model drift: as real-world data distributions change over time, model prediction accuracy gradually declines.61% similarUnverifiedTraditional two-stage methods such as Faster R-CNN rely on a Region Proposal Network (RPN) to first generate candidate boxes, resulting in inference latency of over 100ms59% similarUnverifiedLLM推理延迟分布呈重尾特征，应以P90或P95百分位数而非平均值作为门禁指标57% similarUnverified涌现能力指大模型某些能力在特定参数量阈值后突然出现，由Wei et al. 2022年论文系统记录57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/707119API
curl https://kongchang.com/api/v1/knowledge/claims/707119MCP
get_claim(id=707119)