重新思考细化:无需噪声注入即可纠正生成偏差
Rethinking Refinement: Correcting Generative Bias without Noise Injection
AI总结:
本文提出Bi-stage Flow Refinement框架,通过潜在空间对齐和数据空间细化,无需噪声注入即可有效纠正生成模型的系统性偏差,提升样本保真度和覆盖率。
摘要:
生成模型,包括扩散模型和基于流的模型,常常表现出系统性偏差,这会降低样本质量,尤其是在高维设置中。我们重新审视细化方法,并展示通过后处理过程可以有效纠正偏差,而无需注入噪声或多次重采样采样过程。我们提出了一种基于流匹配的Bi-stage Flow Refinement (BFR)框架,包含两个在不同阶段运行的细化策略:用于近可逆生成器的潜在空间对齐和通过轻量级增强训练的数据空间细化。与之前扰动采样动态的细化器不同,BFR保留了原始ODE轨迹,并对生成样本应用确定性修正。在MNIST、CIFAR-10和FFHQ在256x256分辨率上的实验表明,一致性地提高了保真度和覆盖率;值得注意的是,从基样本(FID 3.95)开始,仅通过一次额外函数评估(1-NFE),潜在空间细化在MNIST上实现了最先进的FID 1.46,同时保持样本多样性。
英文摘要:
Generative models, including diffusion and flow-based models, often exhibit systematic biases that degrade sample quality, particularly in high-dimensional settings. We revisit refinement methods and show that effective bias correction can be achieved as a post-hoc procedure, without noise injection or multi-step resampling of the sampling process. We propose a flow-matching-based \textbf{Bi-stage Flow Refinement (BFR)} framework with two refinement strategies operating at different stages: latent space alignment for approximately invertible generators and data space refinement trained with lightweight augmentations. Unlike previous refiners that perturb sampling dynamics, BFR preserves the original ODE trajectory and applies deterministic corrections to generated samples. Experiments on MNIST, CIFAR-10, and FFHQ at 256x256 resolution demonstrate consistent improvements in fidelity and coverage; notably, starting from base samples with FID 3.95, latent space refinement achieves a \textbf{state-of-the-art} FID of \textbf{1.46} on MNIST using only a single additional function evaluation (1-NFE), while maintaining sample diversity.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。