Curated Synthetic Data Doesn't Have to Collapse: A Theoretical Study of Generative Retraining with Pluralistic Preferences

• URL: http://arxiv.org/abs/2605.07724v1
• Date: Fri, 08 May 2026 13:27:23 GMT
• Title: Curated Synthetic Data Doesn't Have to Collapse: A Theoretical Study of Generative Retraining with Pluralistic Preferences
• Abstract summary: Recursive retraining of generative models poses a critical representation challenge.<n>We show that collapse can be mitigated through curation based on multiple reward functions.
• Score: 13.835275211048113
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recursive retraining of generative models poses a critical representation challenge: when synthetic outputs are curated based on a fixed reward signal, the model tends to collapse onto a narrow set of outputs that over-optimize that objective. Prior work suggests that such collapse is unavoidable without adding real data into the mix. We revisit this conclusion from an alignment perspective and show that collapse can be mitigated through curation based on multiple reward functions. We formalize the dynamics of recursive training under heterogeneous preferences and prove that, under certain conditions, the model converges to a stable distribution that allocates probability mass across competing high-reward regions. The limiting distribution preserves diversity and provably satisfies a weighted Nash bargaining solution, offering a formal interpretation of value aggregation in synthetic retraining loops.

Related papers

• Recursively Trained Diffusion Models: Limiting Collapse Distribution and Spectral Characterization [27.894241484593735]
  Recursive training of generative models can lead to model collapse, a compounding drift away from the true data distribution.<n>We show that even with perfect score estimation and exact sampling, the early stopping of the reverse diffusion drives a progressive drift away from the data distribution.<n>We prove our idealized characterization is robust, with mode-dependent contraction rates that contract high-order errors faster than low-order ones.
  arXiv  Detail & Related papers  (2026-06-11T18:06:26Z)
• Joint Model and Data Sparsification via the Marginal Likelihood [53.29070892356214]
  We propose jointly learning individual feature and sample relevancies, enabling simultaneous model and data sparsification.<n>This symmetric pruning of model and data offers a natural extension that preserves conjugacy.<n> Empirical results across diverse regression tasks affirm that a joint ARD approach consistently yields both sparse and robust prediction models.
  arXiv  Detail & Related papers  (2026-05-28T13:26:53Z)
• Generative Recursive Reasoning [67.22973831501257]
  Generative Recursive reAsoning Models (GRAM) is a framework that turns latent reasoning into probabilistic multi-trajectory.<n>GRAM supports conditional reasoning via $p_(y mid x)$ and, with fixed or absent inputs, unconditional generation via $p_(x)$.
  arXiv  Detail & Related papers  (2026-05-19T05:20:56Z)
• Why Self-Training Helps and Hurts: Denoising vs. Signal Forgetting [6.369253528507392]
  Iterative self-training repeatedly refits a model on pseudo-labels generated by its own predictions.<n>We derive deterministic-equivalent recursions for the prediction risk and effective noise across iterations.
  arXiv  Detail & Related papers  (2026-02-15T07:28:12Z)
• SIGMA: Scalable Spectral Insights for LLM Collapse [51.863164847253366]
  We introduce SIGMA (Spectral Inequalities for Gram Matrix Analysis), a unified framework for model collapse.<n>By utilizing benchmarks that deriving and deterministic bounds on the matrix's spectrum, SIGMA provides a mathematically grounded metric to track the contraction of the representation space.<n>We demonstrate that SIGMA effectively captures the transition towards states, offering both theoretical insights into the mechanics of collapse.
  arXiv  Detail & Related papers  (2026-01-06T19:47:11Z)
• On the Limits of Self-Improving in LLMs and Why AGI, ASI and the Singularity Are Not Near Without Symbolic Model Synthesis [0.01269104766024433]
  We formalise self-training in Large Language Models (LLMs) and Generative AI as a discrete-time dynamical system.<n>We derive two fundamental failure modes: (1) Entropy Decay, where finite sampling effects cause a monotonic loss of distributional diversity (mode collapse), and (2) Variance Amplification, where the loss of external grounding causes the model's representation of truth to drift as a random walk.
  arXiv  Detail & Related papers  (2026-01-05T19:50:49Z)
• Entropy-Reservoir Bregman Projection: An Information-Geometric Unification of Model Collapse [3.533187668612022]
  We present EntropyReser Bregman Projection- ERBP, an information-geometric framework that unifies these phenomena.<n>Our theory yields a necessary condition for collapse, (ii) a sufficient condition that guarantees a non-language entropy floor, and (iii) closed-form rates that depend on sample size.
  arXiv  Detail & Related papers  (2025-12-16T19:50:03Z)
• Latent Iterative Refinement Flow: A Geometric-Constrained Approach for Few-Shot Generation [5.062604189239418]
  We introduce Latent Iterative Refinement Flow (LIRF), a novel approach to few-shot generation.<n>LIRF establishes a stable latent space using an autoencoder trained with our novel textbfmanifold-preservation loss.<n>Within this cycle, candidate samples are refined by a geometric textbfcorrection operator, a provably contractive mapping.
  arXiv  Detail & Related papers  (2025-09-24T08:57:21Z)
• How Bad is Training on Synthetic Data? A Statistical Analysis of Language Model Collapse [9.59833542807268]
  Model collapse occurs when new models are trained on synthetic data generated from previously trained models. We show that model collapse cannot be avoided when training solely on synthetic data. We estimate a maximal amount of synthetic data below which model collapse can eventually be avoided.
  arXiv  Detail & Related papers  (2024-04-07T22:15:13Z)
• Fine-Tuning of Continuous-Time Diffusion Models as Entropy-Regularized Control [54.132297393662654]
  Diffusion models excel at capturing complex data distributions, such as those of natural images and proteins. While diffusion models are trained to represent the distribution in the training dataset, we often are more concerned with other properties, such as the aesthetic quality of the generated images. We present theoretical and empirical evidence that demonstrates our framework is capable of efficiently generating diverse samples with high genuine rewards.
  arXiv  Detail & Related papers  (2024-02-23T08:54:42Z)
• Towards Theoretical Understandings of Self-Consuming Generative Models [56.84592466204185]
  This paper tackles the emerging challenge of training generative models within a self-consuming loop. We construct a theoretical framework to rigorously evaluate how this training procedure impacts the data distributions learned by future models. We present results for kernel density estimation, delivering nuanced insights such as the impact of mixed data training on error propagation.
  arXiv  Detail & Related papers  (2024-02-19T02:08:09Z)
• Time-series Generation by Contrastive Imitation [87.51882102248395]
  We study a generative framework that seeks to combine the strengths of both: Motivated by a moment-matching objective to mitigate compounding error, we optimize a local (but forward-looking) transition policy. At inference, the learned policy serves as the generator for iterative sampling, and the learned energy serves as a trajectory-level measure for evaluating sample quality.
  arXiv  Detail & Related papers  (2023-11-02T16:45:25Z)
• A Twin Neural Model for Uplift [59.38563723706796]
  Uplift is a particular case of conditional treatment effect modeling. We propose a new loss function defined by leveraging a connection with the Bayesian interpretation of the relative risk. We show our proposed method is competitive with the state-of-the-art in simulation setting and on real data from large scale randomized experiments.
  arXiv  Detail & Related papers  (2021-05-11T16:02:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.