Unverified50% confidenceFactExact time
For any compactly supported target distribution, as long as the number of particles is sufficiently large, the MMD at every local minimum in the weighted MMD optimization landscape can be arbitrarily small.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe optimization problem of weighted MMD is non-convex, and when the number of particles is insufficient, gradient flow easily falls into spurious local minima.74% similarUnverifiedIn the field of weighted MMD optimization, previous results on overparameterized benign landscapes were mostly concentrated in neural networks or specific non-convex problems, and the corresponding theory for MMD particle optimization has remained a blank.73% similarUnverifiedWhen the target distribution has finite support, the paper gives an explicit particle count threshold, beyond which no spurious local minima exist and every spurious stationary point is unstable.70% similarUnverifiedTraditional first-order Langevin algorithms (such as projected Langevin Monte Carlo) require that the log-density of the target distribution be differentiable when handling sampling on constrained domains.69% similarUnverifiedMMD（Maximum Mean Discrepancy）是基于再生核希尔伯特空间的非参数分布距离度量，相比FID对小样本场景更友好、计算开销更低68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1037443API
curl https://kongchang.com/api/v1/knowledge/claims/1037443MCP
get_claim(id=1037443)