Elucidating the SNR-t Bias of Diffusion Probabilistic Models

2026-04-17

AI Summary

This work identifies and systematically analyzes a critical misalignment between signal-to-noise ratio (SNR) and time steps during the inference of diffusion probabilistic models, which degrades generation quality. To address this SNR–t discrepancy, the authors propose a frequency-aware differential correction method that separately calibrates high- and low-frequency components in the denoising process, effectively mitigating the temporal misalignment between SNR and sampling steps. The approach introduces negligible computational overhead while consistently enhancing generation performance across a wide range of state-of-the-art diffusion models—including IDDPM, ADM, DDIM, EDM, PFGM++, and FLUX—on diverse datasets spanning multiple resolutions.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Diffusion Probabilistic Models have demonstrated remarkable performance across a wide range of generative tasks. However, we have observed that these models often suffer from a Signal-to-Noise Ratio-timestep (SNR-t) bias. This bias refers to the misalignment between the SNR of the denoising sample and its corresponding timestep during the inference phase. Specifically, during training, the SNR of a sample is strictly coupled with its timestep. However, this correspondence is disrupted during inference, leading to error accumulation and impairing the generation quality. We provide comprehensive empirical evidence and theoretical analysis to substantiate this phenomenon and propose a simple yet effective differential correction method to mitigate the SNR-t bias. Recognizing that diffusion models typically reconstruct low-frequency components before focusing on high-frequency details during the reverse denoising process, we decompose samples into various frequency components and apply differential correction to each component individually. Extensive experiments show that our approach significantly improves the generation quality of various diffusion models (IDDPM, ADM, DDIM, A-DPM, EA-DPM, EDM, PFGM++, and FLUX) on datasets of various resolutions with negligible computational overhead. The code is at https://github.com/AMAP-ML/DCW.

Problem

• Diffusion Probabilistic Models
• SNR-t bias
• Signal-to-Noise Ratio
• timestep misalignment
• generation quality

Innovation

• SNR-t bias
• diffusion probabilistic models
• differential correction
• frequency decomposition
• generative modeling