#machine learning
Apr 2026
Understanding Latent Diffusability via Fisher Geometry
It is shown that isometric embeddings preserve intrinsic FI and establish quantitative intrinsic-FI bounds for a broader class of bi-Lipschitz encoders with controlled weak volume distortion, whereas FIR is governed by the interplay between encoder and data geometries.
Jing Gu, M. Mardani, Wonjun Lee et al.
· arXiv.org · 0 citations