English

Explicit Density Approximation for Neural Implicit Samplers Using a Bernstein-Based Convex Divergence

Machine Learning 2025-11-07 v2 Artificial Intelligence Probability Machine Learning

Abstract

Rank-based statistical metrics, such as the invariant statistical loss (ISL), have recently emerged as robust and practically effective tools for training implicit generative models. In this work, we introduce dual-ISL, a novel likelihood-free objective for training implicit generative models that interchanges the roles of the target and model distributions in the ISL framework, yielding a convex optimization problem in the space of model densities. We prove that the resulting rank-based discrepancy dKd_K is i) continuous under weak convergence and with respect to the L1L^1 norm, and ii) convex in its first argument-properties not shared by classical divergences such as KL or Wasserstein distances. Building on this, we develop a theoretical framework that interprets dKd_K as an L2L^2-projection of the density ratio q=p/p~q = p/\tilde p onto a Bernstein polynomial basis, from which we derive exact bounds on the truncation error, precise convergence rates, and a closed-form expression for the truncated density approximation. We further extend our analysis to the multivariate setting via random one-dimensional projections, defining a sliced dual-ISL divergence that retains both convexity and continuity. We empirically show that these theoretical advantages translate into practical ones. Specifically, across several benchmarks dual-ISL converges more rapidly, delivers markedly smoother and more stable training, and more effectively prevents mode collapse than classical ISL and other leading implicit generative methods-while also providing an explicit density approximation.

Keywords

Cite

@article{arxiv.2506.04700,
  title  = {Explicit Density Approximation for Neural Implicit Samplers Using a Bernstein-Based Convex Divergence},
  author = {José Manuel de Frutos and Manuel A. Vázquez and Pablo M. Olmos and Joaquín Míguez},
  journal= {arXiv preprint arXiv:2506.04700},
  year   = {2025}
}
R2 v1 2026-07-01T03:00:47.144Z