Skip to content

Let It Go or Learn to Self-Correct: Continuous Diffusion for Constrained Discrete Tasks

Sep 2026 · arXiv.org · Vol abs/2609.09009 · 0 citations · 58 references
Computer Science

TL;DR

The results suggest that continuous diffusion models can learn nontrivial global constraints, but discrete reasoning tasks require better alignment between training and inference: either through samplers that reduce commitment to early decisions, or through training that teaches the model to correct its own inference-time errors.

Abstract

Denoising Diffusion Probabilistic Models (DDPMs) generate samples by starting from noise and repeatedly denoising while keeping each update close to the current noisy state. This behavior is effective in many continuous domains, but its role is less clear for globally constrained discrete tasks, such as Sudoku, graph connectivity, Latin squares, and N-queens. In such settings, early discrete errors can be difficult to undo. As a result, standard diffusion sampling may preserve early mistakes, even when the model's clean predictions are informative. We compare standard samplers to sampling directly from the model's clean prediction. Without retraining, this single change improves Sudoku validity from 31% to 95%, with consistent gains across the other discrete tasks. We hypothesize that staying close to the current noisy state is harmful because the reverse trajectory can drift off the forward noising distribution the model was trained on. To reduce this train-test mismatch, we further introduce self-correction training, which exposes the model to its own predictions, improving robustness to errors that arise during inference. This substantially improves the performance of standard samplers. Our results suggest that continuous diffusion models can learn nontrivial global constraints, but discrete reasoning tasks require better alignment between training and inference: either through samplers that reduce commitment to early decisions, or through training that teaches the model to correct its own inference-time errors.

View source

Similar papers

#data science Sep 2026

FluxLite: Inference-Time Proposal Control for Discrete Diffusion Models

FluxLite is introduced, a lightweight, training-free proposal-control framework for discrete diffusion, identifying a tilted-path coverage factor that governs robustness to score error, together with finite-particle convergence for a fixed controlled Feynman-Kac recursion.

Yinuo Ren, Haoxuan Chen, Grant M. Rotskoff et al. · 1 citation
#machine learning Preprint Oct 2026

Score Mismatch and Set-Constrained Sampling in Zero-Shot Conditional Discrete Diffusion

Discrete diffusion models have become a powerful framework for modeling complex discrete distributions across a range of applications. While theoretical convergence guarantees depend on consistency between the target and training distributions, such a case rarely holds in practice, making score mismatch a central chall...

Ren-Xiang Huang, Yu-Chen Liang, Ying-Bin Liang et al. · 0 citations

CAST: Causal Advantage-Structured Training with Spatially Grounded Compositional Rewards for Diffusion Models

CAST (Causal Advantage-Structured Training), an RL fine-tuning method for pretrained DMs, which identifies the denoising step at which each model fixes the objects and their spatial arrangement in the image and uses that timing to set the SDE window and uses them to spatially weight the SDE policy objective.

Shu Yu, Chao-Chao Lu · 0 citations
Sep 2026

ExploreNet: Learning Where to Explore in Diffusion GRPO

Across group-relative diffusion RL experiments, it is found that exploration is learnable, the shape of the exploration distribution outweighs its magnitude, and rollout quality is more effective than rollout quantity.

S. Li, Xiao-Chuang Han, Y. Tsvetkov et al. · 0 citations
#artificial intelligence Preprint Oct 2026

Correction Training for Diffusion Models

The method substantially improves upon previously reported FID scores for unguided generation and, notably, unconditional generation, where producing high-quality samples without guidance has been particularly challenging.

Donghoon Ahn, Run-Qian Wang, Alexei A. Efros et al. · 0 citations
#machine learning Preprint Oct 2026

Ambient Discrete Diffusion: Using the Wrong Data at the Right Time for Data Efficient Learning

We introduce RefineMix, a framework for training discrete diffusion models under severe data scarcity, a common constraint in scientific applications. RefineMix uses out-of-distribution data at selected diffusion times to improve generalization without biasing the sampling distribution. Although this strategy has been...

Julian Kleutgens, M. Tec, Claudio Battiloro et al. · 0 citations

Related blog posts

MIT News · Artificial Intelligence Sep 29, 2026

Who we become when we talk to machines

Professor Sherry Turkle’s new book, “Artificial Intimacy,” offers a withering critique of chatbots and the antisocial dynamics she believes they encourage.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.