The paper examines generative models that are trained through self‑consuming loops where users curate preferred samples. It shows that when training relies solely on user‑curated synthetic data, initial biases are amplified, leading the system to converge to a single dominant outcome. Introducing reference data at a sufficient rate changes the dynamics, producing a unique globally attracting equilibrium, and the authors propose an algorithm to select reference distributions and mixing weights to steer the system toward desirable equilibria while reducing data collection costs.
arXiv:2609.15598v1 Announce Type: cross Abstract: Generative recommendation has emerged as a promising end-to-end paradigm for personalized recommendation. However, user preferences continuously evol...
arXiv:2605. 07724v2 Announce Type: replace-cross 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.
The paper examines how multiple generative models can form a networked self‑consuming loop, where each model consumes synthetic data produced by others. It introduces a graph‑based theoretical framework that models these interactions as directed, weighted edges and uses it to analyze long‑term retraining dynamics, identifying conditions for convergence and the nature of fixed points. The study also explores how real‑data access, cross‑model consumption, and graph structure influence the system’s stability and diversity.
arXiv:2601. 09974v2 Announce Type: replace Abstract: Personalizing Large Language Models typically relies on static retrieval or one-time adaptation, assuming user preferences remain invariant over time.
arXiv:2606. 05264v1 Announce Type: new Abstract: Training robust multivariate time series forecasting models requires large, diverse corpora, yet many real-world domains provide only a handful of observed sequences.
arXiv:2502. 04646v2 Announce Type: replace-cross Abstract: Weighted sampling -- sampling from a probability density function (PDF) proportional to the product of a base PDF and a weight function -- is a fundamental technique with wide-ranging applications in variance reduction, biased sampling, data augmentation, and more.
arXiv:2510. 16657v3 Announce Type: replace-cross Abstract: Synthetic data has been increasingly used to train frontier generative models.
The paper introduces Reflection-Aware GRPO (RA‑GRPO), a reinforcement‑learning framework that aligns diffusion generative models with human preferences. It uses Diffusion Reflection to correct intermediate sampling paths by reversing the diffusion process, and Counterfactual Path Synthesis to embed these corrected trajectories into the policy, avoiding extra inference cost. Experiments on text‑to‑image and text‑to‑video models show RA‑GRPO outperforms existing methods, reducing reward hacking and improving generalization while remaining architecture‑agnostic.
The paper compares two strategies for contextual decision making: predict‑then‑optimize (PtO), which learns the conditional expected objective, and generate‑then‑optimize (GtO), which learns conditional uncertainty and optimizes over generated scenarios. It analyzes how the choice of learning target influences statistical accuracy and optimization tractability, showing that smoothness can compensate for GtO’s higher dimensionality. The study also discusses tradeoffs in convexity preservation, scenario‑wise convexity, and model flexibility, supported by synthetic experiments and a real case study.
The paper presents a scalable method for learning dataweights that optimizes training-example importance using metagradients, guided by a small high‑quality target dataset. It introduces parametric weighting rules that generalize to new and streaming data and can transfer from smaller to larger models. Applied to reward modeling, the learned weights improve held‑out performance, adapt models to specialized domains, and enable human‑synthetic supervision mixtures to outperform either source alone.
arXiv:2609.37800v1 Announce Type: cross Abstract: Many recommender services repeatedly encounter cold-start cohorts, where new users arrive with little or no interaction history. This creates two cha...
arXiv:2406.02447v5 Announce Type: replace Abstract: Federated Learning (FL) aims at unburdening the training of deep models by distributing computation across multiple devices (clients) while safegua...