Neural Manifold Ordinary Differential Equations
To better conform to data geometry, recent deep generative modelling techniques adapt Euclidean constructions to non-Euclidean spaces. In this paper, we study normalizing flows on manifolds. Previous work has developed flow models for specific cases; however, these advancements hand craft layers on a manifold-by-manifold basis, restricting generality and inducing cumbersome design constraints. We overcome these issues by introducing Neural Manifold Ordinary Differential Equations, a manifold generalization of Neural ODEs, which enables the construction of Manifold Continuous Normalizing Flows (MCNFs). MCNFs require only local geometry (therefore generalizing to arbitrary manifolds) and compute probabilities with continuous change of variables (allowing for a simple and expressive flow construction). We find that leveraging continuous manifold dynamics produces a marked improvement for both density estimation and downstream tasks.
Neural Manifold Ordinary Differential Equations Aaron Lou*, Derek Lim*, Isay Katsman*, Leo Huang*, Qingxuan Jiang Cornell University {al968, dl772, isk22, ah839, qj46}@cornell.edu arXiv:2006.10254v1 [stat.ML] 18 Jun 2020 Ser-Nam Lim Christopher De Sa…
saved by
related reading
- Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gaugesarxiv.org
- Modular Manifolds - Thinking Machines Labthinkingmachines.ai
- Normalizing Flows for Probabilistic Modeling and Inferencealphaxiv.org
- Flow Matching for Generative Modelingalphaxiv.org
- Normalizing flowsbehindml.com
- 1806.07366arxiv.org
- Neural Ordinary Differential Equationsarxiv.org
- Neural Networks, Manifolds, and Topology -- colah's blogcolah.github.io
- Learning the integral of a diffusion model – Sander Dielemansander.ai
- Denoising Density Estimatorsarxiv.org
- What are Diffusion Models? | Lil'Loglilianweng.github.io
- An Introduction to Flow Matching and Diffusion Modelsdiffusion.csail.mit.edu