Exposure Bias Can Alleviate Itself via Directional and Frequency Rectification in Flow Matching
2026-06-26 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors address a problem in Flow Matching (FM) models where the training setup doesn't perfectly match how the model is used during inference, causing errors called exposure bias. They propose a method called DEFAR that uses the bias itself as a guide to fix these errors dynamically during training. DEFAR has two parts: one helps the model learn to correct itself when it drifts off track, and the other fixes missing signal patterns that cause bias. Tests on popular image datasets show that their approach improves generation quality and is robust across different settings.
Flow MatchingExposure BiasGenerative ModelsInference DriftLow-frequency ComponentsSelf-rectificationImage GenerationCIFAR-10CelebAImageNet
Authors
Guanbo Huang, Jingjia Mao, Fanding Huang, Fengkai Liu, Xiangyang Luo, Yaoyuan Liang, Jiasheng Lu, Xiaoe Wang, Pei Liu, Ruiliu Fu, Ruqi Huang, Shao-Lun Huang
Abstract
Flow Matching (FM) has achieved remarkable generative performance, yet it suffers from exposure bias due to discrepancies between training and inference. Existing mitigation strategies typically rely on static constraints or external heuristics. In this work, we propose that exposure bias itself inherently contains dynamic signals that can guide its own rectification. To leverage this, we introduce DEFAR (DirEctional-Frequency Adaptive Rectification). This framework simulates the single-step inference process during training to identify exposure bias. It utilizes directional and frequency-adaptive feedback signals from the bias itself to enhance the model's bias tolerance. It consists of two key components: (1) Anti-Drift Rectification (ADR). ADR treats inference-time drift as a signal to learn the direction to steer deviated states back toward the target. ADR endows the model with intrinsic active self-rectification capabilities; (2) Frequency Compensation (FC). Empirically, we observe that accumulated bias often stems from a lack of low-frequency components in high-noise stages, and exposure bias carries the missing frequency. FC leverages the bias itself as a self-feedback weighting factor to reinforce the missing frequency components. Experiments on CIFAR-10, CelebA-64, and ImageNet-256/512 show that DEFAR outperforms prior baselines and further demonstrates favorable scalability, compatibility, and inference robustness.