Unverified50% confidenceFactExact time
流形假设由Bengio、LeCun等深度学习先驱提出，认为高维真实数据近似分布在低维光滑流形上
1
Sources
50%
Confidence
Long-term
Relevance
7/14/2026
First Seen
Sources
引导式生成模型：精准估算极端事件概率的新方法
Related Claims
Unverified深度生成模型通过学习高维数据中的低维流形结构来规避维度灾难69% similarUnverifiedThe poor scalability of statistical modeling methods on high-dimensional data is the core reason for the gradual rise of deep learning methods.69% similarUnverifiedDeep learning relies on linear algebra, calculus, probability theory, and optimization theory.67% similarUnverifiedGrinsztajn等人2022年在NeurIPS发表的基准研究证明，在中等规模表格数据集上，GBDT类模型持续优于深度学习方法67% similarUnverifiedDeepPose first introduced deep learning into the field of pose estimation in 2014.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/510572API
curl https://kongchang.com/api/v1/knowledge/claims/510572MCP
get_claim(id=510572)