Generative model for optimal density estimation on unknown manifold

TL;DR

This paper introduces a generative model that adaptively estimates distributions on unknown low-dimensional manifolds, achieving optimal convergence rates and outperforming existing models in various datasets.

Contribution

The proposed model leverages geometric insights to adapt to the data's manifold structure, ensuring minimax-optimal convergence for a broad class of metrics.

Findings

Achieves minimax-optimal convergence rates for manifold-supported distributions.

Supports a wide range of Hölder IPMs with b3 a9 1.

Demonstrates superior performance on synthetic and real datasets.

Abstract

We propose a generative model that achieves minimax-optimal convergence rates for estimating probability distributions supported on unknown low-dimensional manifolds. Building on Fefferman's solution to the geometric Whitney problem, our estimator is itself supported on a submanifold that matches the regularity of the data's support. This geometric adaptation enables the estimator to be simultaneously minimax-optimal for all \( \gamma \)-H\"older Integral Probability Metrics (IPMs) with \( \gamma \geq 1 \). We validate our approach through experiments on synthetic and real datasets, demonstrating competitive or superior performance compared to Wasserstein GAN and score-based generative models.