{"ID":2823310,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2601.00781","arxiv_id":"2601.00781","title":"Categorical Reparameterization with Denoising Diffusion models","abstract":"Learning models with categorical variables requires optimizing expectations over discrete distributions, a setting in which stochastic gradient-based optimization is challenging due to the non-differentiability of categorical sampling. A common workaround is to replace the discrete distribution with a continuous relaxation, yielding a smooth surrogate that admits reparameterized gradient estimates via the reparameterization trick. Building on this idea, we introduce ReDGE, a novel and efficient diffusion-based soft reparameterization method for categorical distributions. Our approach defines a flexible class of gradient estimators that includes the Straight-Through estimator as a special case. Experiments spanning latent variable models and inference-time reward guidance in discrete diffusion models demonstrate that ReDGE consistently matches or outperforms existing gradient-based methods. The code will be made available at https://github.com/samsongourevitch/redge.","short_abstract":"Learning models with categorical variables requires optimizing expectations over discrete distributions, a setting in which stochastic gradient-based optimization is challenging due to the non-differentiability of categorical sampling. A common workaround is to replace the discrete distribution with a continuous relaxa...","url_abs":"https://arxiv.org/abs/2601.00781","url_pdf":"https://arxiv.org/pdf/2601.00781v2","authors":"[\"Samson Gourevitch\",\"Alain Durmus\",\"Eric Moulines\",\"Jimmy Olsson\",\"Yazid Janati\"]","published":"2026-01-02T18:30:05Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"stat.ML\"]","methods":"[\"Diffusion Model\"]","has_code":false,"code_links":[{"ID":605485,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2823310,"paper_url":"https://arxiv.org/abs/2601.00781","paper_title":"Categorical Reparameterization with Denoising Diffusion models","repo_url":"https://github.com/samsongourevitch/redge","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}