Masked Diffusion Models are Secretly Time-Agnostic Masked Models and Exploit Inaccurate Categorical Sampling

• URL: http://arxiv.org/abs/2409.02908v3
• Date: Fri, 25 Oct 2024 02:54:34 GMT
• Title: Masked Diffusion Models are Secretly Time-Agnostic Masked Models and Exploit Inaccurate Categorical Sampling
• Authors: Kaiwen Zheng, Yongxin Chen, Hanzi Mao, Ming-Yu Liu, Jun Zhu, Qinsheng Zhang,
• Abstract summary: Masked diffusion models (MDMs) have emerged as a popular research topic for generative modeling of discrete data. We show that both training and sampling of MDMs are theoretically free from the time variable. We identify, for the first time, an underlying numerical issue, even with the commonly used 32-bit floating-point precision.
• Score: 47.82616476928464
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Masked diffusion models (MDMs) have emerged as a popular research topic for generative modeling of discrete data, thanks to their superior performance over other discrete diffusion models, and are rivaling the auto-regressive models (ARMs) for language modeling tasks. The recent effort in simplifying the masked diffusion framework further leads to alignment with continuous-space diffusion models and more principled training and sampling recipes. In this paper, however, we reveal that both training and sampling of MDMs are theoretically free from the time variable, arguably the key signature of diffusion models, and are instead equivalent to masked models. The connection on the sampling aspect is drawn by our proposed first-hitting sampler (FHS). Specifically, we show that the FHS is theoretically equivalent to MDMs' original generation process while significantly alleviating the time-consuming categorical sampling and achieving a 20$\times$ speedup. In addition, our investigation raises doubts about whether MDMs can truly beat ARMs in text generation. We identify, for the first time, an underlying numerical issue, even with the commonly used 32-bit floating-point precision, which results in inaccurate categorical sampling. We show that it lowers the effective temperature both theoretically and empirically, and the resulting decrease in token diversity makes previous evaluations, which assess the generation quality solely through the incomplete generative perplexity metric, somewhat unfair.

Related papers

• Table-to-Text Generation with Pretrained Diffusion Models [0.0]
  Diffusion models have demonstrated significant potential in achieving state-of-the-art performance across various text generation tasks. We investigate their application to the table-to-text problem by adapting the diffusion model to the task and conducting an in-depth analysis. Our findings reveal that diffusion models achieve comparable results in the table-to-text domain.
  arXiv  Detail & Related papers  (2024-09-10T15:36:53Z)
• Provable Statistical Rates for Consistency Diffusion Models [87.28777947976573]
  Despite the state-of-the-art performance, diffusion models are known for their slow sample generation due to the extensive number of steps involved. This paper contributes towards the first statistical theory for consistency models, formulating their training as a distribution discrepancy minimization problem.
  arXiv  Detail & Related papers  (2024-06-23T20:34:18Z)
• Boosting Diffusion Models with Moving Average Sampling in Frequency Domain [101.43824674873508]
  Diffusion models rely on the current sample to denoise the next one, possibly resulting in denoising instability. In this paper, we reinterpret the iterative denoising process as model optimization and leverage a moving average mechanism to ensemble all the prior samples. We name the complete approach "Moving Average Sampling in Frequency domain (MASF)"
  arXiv  Detail & Related papers  (2024-03-26T16:57:55Z)
• Uncertainty-aware Surrogate Models for Airfoil Flow Simulations with Denoising Diffusion Probabilistic Models [26.178192913986344]
  We make a first attempt to use denoising diffusion probabilistic models (DDPMs) to train an uncertainty-aware surrogate model for turbulence simulations. Our results show DDPMs can successfully capture the whole distribution of solutions and, as a consequence, accurately estimate the uncertainty of the simulations. We also evaluate an emerging generative modeling variant, flow matching, in comparison to regular diffusion models.
  arXiv  Detail & Related papers  (2023-12-08T19:04:17Z)
• CamoDiffusion: Camouflaged Object Detection via Conditional Diffusion Models [72.93652777646233]
  Camouflaged Object Detection (COD) is a challenging task in computer vision due to the high similarity between camouflaged objects and their surroundings. We propose a new paradigm that treats COD as a conditional mask-generation task leveraging diffusion models. Our method, dubbed CamoDiffusion, employs the denoising process of diffusion models to iteratively reduce the noise of the mask.
  arXiv  Detail & Related papers  (2023-05-29T07:49:44Z)
• Membership Inference Attacks against Diffusion Models [0.0]
  Diffusion models have attracted attention in recent years as innovative generative models. We investigate whether a diffusion model is resistant to a membership inference attack.
  arXiv  Detail & Related papers  (2023-02-07T05:20:20Z)
• Fast Inference in Denoising Diffusion Models via MMD Finetuning [23.779985842891705]
  We present MMD-DDM, a novel method for fast sampling of diffusion models. Our approach is based on the idea of using the Maximum Mean Discrepancy (MMD) to finetune the learned distribution with a given budget of timesteps. Our findings show that the proposed method is able to produce high-quality samples in a fraction of the time required by widely-used diffusion models.
  arXiv  Detail & Related papers  (2023-01-19T09:48:07Z)
• How Much is Enough? A Study on Diffusion Times in Score-based Generative Models [76.76860707897413]
  Current best practice advocates for a large T to ensure that the forward dynamics brings the diffusion sufficiently close to a known and simple noise distribution. We show how an auxiliary model can be used to bridge the gap between the ideal and the simulated forward dynamics, followed by a standard reverse diffusion process.
  arXiv  Detail & Related papers  (2022-06-10T15:09:46Z)
• Denoising Diffusion Implicit Models [117.03720513930335]
  We present denoising diffusion implicit models (DDIMs) for iterative implicit probabilistic models with the same training procedure as DDPMs. DDIMs can produce high quality samples $10 times$ to $50 times$ faster in terms of wall-clock time compared to DDPMs.
  arXiv  Detail & Related papers  (2020-10-06T06:15:51Z)

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.