Equivariant Manifold Flows
Authors:
Isay Katsman,
Aaron Lou,
Derek Lim,
Qingxuan Jiang,
Ser-Nam Lim,
Christopher De Sa
Abstract:
Tractably modelling distributions over manifolds has long been an important goal in the natural sciences. Recent work has focused on develo** general machine learning models to learn such distributions. However, for many applications these distributions must respect manifold symmetries -- a trait which most previous models disregard. In this paper, we lay the theoretical foundations for learning…
▽ More
Tractably modelling distributions over manifolds has long been an important goal in the natural sciences. Recent work has focused on develo** general machine learning models to learn such distributions. However, for many applications these distributions must respect manifold symmetries -- a trait which most previous models disregard. In this paper, we lay the theoretical foundations for learning symmetry-invariant distributions on arbitrary manifolds via equivariant manifold flows. We demonstrate the utility of our approach by using it to learn gauge invariant densities over $SU(n)$ in the context of quantum field theory.
△ Less
Submitted 27 January, 2022; v1 submitted 18 July, 2021;
originally announced July 2021.
Neural Manifold Ordinary Differential Equations
Authors:
Aaron Lou,
Derek Lim,
Isay Katsman,
Leo Huang,
Qingxuan Jiang,
Ser-Nam Lim,
Christopher De Sa
Abstract:
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…
▽ More
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.
△ Less
Submitted 17 June, 2020;
originally announced June 2020.
