提出无架构限制的局部后验崩溃控制方法，提升VAE生成多样性。
Toward Architecture-Agnostic Local Control of Posterior Collapse in VAEs
- 基于注入函数性质设计可适配任意架构的隐变量重构损失
- 在多个数据集上有效缓解后验崩溃，生成样本更丰富多样
- 适合关注生成质量且不愿受限于网络结构的研究者
变分自编码器（VAE）作为主流生成模型，常面临后验崩溃问题，导致生成样本多样性下降。现有方法多通过调控正则化损失来缓解，但重建与正则化之间权衡不佳。部分工作虽引入隐变量可识别性以避免后验崩溃，却需对网络结构施加约束。为此，本文定义了局部后验崩溃，强调数据空间中每个样本点的重要性，并提出一种受注入函数与复合函数数学性质启发的隐变量重构（Latent Reconstruction, LR）损失。该方法无需特定网络架构限制，可在多种数据集（如MNIST、FashionMNIST、Omniglot、CelebA、FFHQ）上有效控制后验崩溃。
原文摘要 · Abstract (English)
Variational autoencoders (VAEs), one of the most widely used generative models, are known to suffer from posterior collapse, a phenomenon that reduces the diversity of generated samples. To avoid posterior collapse, many prior works have tried to control the influence of regularization loss. However, the trade-off between reconstruction and regularization is not satisfactory. For this reason, several methods have been proposed to guarantee latent identifiability, which is the key to avoiding posterior collapse. However, they require structural constraints on the network architecture. For further clarification, we define local posterior collapse to reflect the importance of individual sample points in the data space and to relax the network constraint. Then, we propose Latent Reconstruction(LR) loss, which is inspired by mathematical properties of injective and composite functions, to control posterior collapse without restriction to a specific architecture. We experimentally evaluate our approach, which controls posterior collapse on varied datasets such as MNIST, fashionMNIST, Omniglot, CelebA, and FFHQ.
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。