English

Convergence and Stability Analysis of Self-Consuming Generative Models with Heterogeneous Human Curation

Machine Learning 2025-11-14 v2 Machine Learning

Abstract

Self-consuming generative models have received significant attention over the last few years. In this paper, we study a self-consuming generative model with heterogeneous preferences that is a generalization of the model in Ferbach et al. (2024). The model is retrained round by round using real data and its previous-round synthetic outputs. The asymptotic behavior of the retraining dynamics is investigated across four regimes using different techniques including the nonlinear Perron--Frobenius theory. Our analyses improve upon that of Ferbach et al. (2024) and provide convergence results in settings where the well-known Banach contraction mapping arguments do not apply. Stability and non-stability results regarding the retraining dynamics are also given.

Keywords

Cite

@article{arxiv.2511.09002,
  title  = {Convergence and Stability Analysis of Self-Consuming Generative Models with Heterogeneous Human Curation},
  author = {Hongru Zhao and Jinwen Fu and Tuan Pham},
  journal= {arXiv preprint arXiv:2511.09002},
  year   = {2025}
}

Comments

42 pages, 2 tables

R2 v1 2026-07-01T07:33:25.441Z