通过动态缓存加速频域扩散模型推理，兼顾速度与生成质量。
Accelerating Frequency Domain Diffusion Models with Error-Feedback Event-Driven Caching
- 基于误差反馈和事件驱动机制，智能缓存高频特征。
- 在5个数据集上实现约2.2倍加速，保持样本质量不变。
- 适合需要高效时序生成的场景，如金融、医疗预测。
扩散模型在时序生成中表现优异，但推理速度慢限制了实际应用。本文提出E²-CRF（误差反馈事件驱动累积残差特征缓存）方法，加速频域扩散模型。利用信号能量集中在低频的谱局部性，以及频域镜像对称性（有效维度减半）两个结构特性，构建闭环误差反馈系统，自适应地在扩散步骤间缓存Transformer的键值（KV）特征。通过事件驱动的残差动态触发重计算，仅对高能量或快速变化的令牌进行重算，稳定高频成分则复用缓存。该方法在5个数据集上实现约2.2倍加速，同时保持生成质量。缓存策略自然契合扩散过程从结构到细节的演化规律。附录中在标准正则性假设下给出了充分条件误差界和复杂度分析，并有实证验证。代码已开源于https://github.com/NoakLiu/FastFourierDiffusion，亦集成于https://github.com/NoakLiu/FastCache-xDiT。
原文摘要 · Abstract (English)
Diffusion models achieve remarkable success in time series generation. However, slow inference limits their practical deployment. We propose E$^2$-CRF (Error-Feedback Event-Driven Cumulative Residual Feature caching) to accelerate frequency domain diffusion models. Our method exploits two structural properties: (1) spectral localization, where signal energy concentrates in low frequencies, and (2) mirror symmetry, which halves the effective frequency dimension. E$^2$-CRF uses a closed-loop error-feedback system that adaptively caches transformer KV features across diffusion steps. We trigger recomputation using event-driven residual dynamics instead of fixed schedules. Our method selectively recomputes high-energy or rapidly-changing tokens while reusing cached features for stable high-frequency components. E$^2$-CRF achieves ~2.2 speedup while maintaining sample quality. We demonstrate effectiveness on 5 datasets. Our caching strategy naturally aligns with the diffusion process's structure-to-detail progression. We include sufficient-condition error and complexity bounds under standard regularity assumptions (Appendix), alongside empirical validation. Our code is available at https://github.com/NoakLiu/FastFourierDiffusion and is also integrated in https://github.com/NoakLiu/FastCache-xDiT.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。