Manifold Learning on Sphere N=2,000, σ=0.05 benchmark leaderboard · SOTA2 Research

Manifold Learning on Sphere N=2,000, σ=0.05

99.6TrustworthinessUMAP

Evaluation Results

Method	Links
UMAP2026.07		99.6	96.3	0.679	0.686
t-SNE2026.07		99.5	96.6	0.682	0.692
EntroPath2026.07		90.4	97.7	0.762	0.761
Shortest Path2026.07		87.6	98	0.776	0.768
MDS2026.07		87.5	97.9	0.771	0.766
PHATE2026.07		86.1	98.1	0.766	0.757
HeatGeo2026.07		86	98.2	0.766	0.759
DTNE2026.07		86	97.9	0.768	0.759
Laplacian Eigenmaps2026.07		85.6	98.7	0.752	0.747
Diffusion Maps2026.07		85.2	98.8	0.753	0.748
PCA2026.07		84.8	98.8	0.755	0.751
Isomap2026.07		84.8	98.9	0.757	0.752