SimFlow：简化且端到端训练的潜在规范化流
SimFlow: Simplified and End-to-End Training of Latent Normalizing Flows
AI总结：
SimFlow通过固定VAE方差简化训练流程，提升生成质量，在ImageNet生成任务中取得新高。
摘要：
规范化流（NFs）学习数据与高斯分布之间的可逆映射。先前的工作通常面临两个限制。首先，它们在训练样本或VAE潜在编码器上添加随机噪声作为数据增强，引入了复杂的流程，包括额外的噪声和去噪声步骤。其次，它们使用预训练和冻结的VAE编码器，导致重建和生成质量不理想。在本文中，我们发现这两个问题可以通过非常简单的方法解决：仅仅将方差（否则由VAE编码器预测）固定为一个常数（例如0.5）。一方面，这种方法允许编码器输出更广的令牌分布，解码器从增强的令牌分布中学习重建干净的图像，避免额外的噪声或去噪声设计。另一方面，固定方差简化了VAE证据下界，使其在联合训练NF和VAE时更加稳定。在ImageNet 256×256生成任务中，我们的模型SimFlow获得了2.15的gFID分数，优于最先进的方法STARFlow（gFID 2.40）。此外，SimFlow可以无缝集成到端到端表示对齐（REPA-E）方法中，并实现了1.91的改进gFID分数，在NFs中设定了新的状态。
英文摘要：
Normalizing Flows (NFs) learn invertible mappings between the data and a Gaussian distribution. Prior works usually suffer from two limitations. First, they add random noise to training samples or VAE latents as data augmentation, introducing complex pipelines including extra noising and denoising steps. Second, they use a pretrained and frozen VAE encoder, resulting in suboptimal reconstruction and generation quality. In this paper, we find that the two issues can be solved in a very simple way: just fixing the variance (which would otherwise be predicted by the VAE encoder) to a constant (e.g., 0.5). On the one hand, this method allows the encoder to output a broader distribution of tokens and the decoder to learn to reconstruct clean images from the augmented token distribution, avoiding additional noise or denoising design. On the other hand, fixed variance simplifies the VAE evidence lower bound, making it stable to train an NF with a VAE jointly. On the ImageNet $256 \times 256$ generation task, our model SimFlow obtains a gFID score of 2.15, outperforming the state-of-the-art method STARFlow (gFID 2.40). Moreover, SimFlow can be seamlessly integrated with the end-to-end representation alignment (REPA-E) method and achieves an improved gFID of 1.91, setting a new state of the art among NFs.
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。
