用信息瓶颈优化GAN，实现可解释的解耦表征学习
IB-GAN: Disentangled Representation Learning with Information Bottleneck Generative Adversarial Networks
- 在生成器中间层施加信息瓶颈，约束输入与输出间互信息
- 在dSprites数据集上解耦得分媲美最优VAE，在CelebA上生成质量更优
- 适合需要可控生成和可解释性表示的研究者使用
我们提出一种基于GAN的无监督解耦表征学习新模型。该模型通过将信息瓶颈（IB）框架引入GAN优化过程而得名IB-GAN。其架构与InfoGAN部分相似，但关键不同在于：利用生成器的中间层来约束输入与生成输出之间的互信息。该中间随机层可作为可学习的潜在分布，与生成器端到端联合训练。结果，IB-GAN的生成器能以解耦且可解释的方式使用潜在空间。在dSprites和Color-dSprites数据集上的实验表明，IB-GAN的解耦得分达到与最先进η-VAE相当的水平，优于InfoGAN。此外，在CelebA和3D Chairs数据集上，其生成样本的视觉质量和多样性也优于η-VAE和InfoGAN，FID得分表现更优。
原文摘要 · Abstract (English)
We propose a new GAN-based unsupervised model for disentangled representation learning. The new model is discovered in an attempt to utilize the Information Bottleneck (IB) framework to the optimization of GAN, thereby named IB-GAN. The architecture of IB-GAN is partially similar to that of InfoGAN but has a critical difference; an intermediate layer of the generator is leveraged to constrain the mutual information between the input and the generated output. The intermediate stochastic layer can serve as a learnable latent distribution that is trained with the generator jointly in an end-to-end fashion. As a result, the generator of IB-GAN can harness the latent space in a disentangled and interpretable manner. With the experiments on dSprites and Color-dSprites dataset, we demonstrate that IB-GAN achieves competitive disentanglement scores to those of state-of-the-art \b{eta}-VAEs and outperforms InfoGAN. Moreover, the visual quality and the diversity of samples generated by IB-GAN are often better than those by \b{eta}-VAEs and Info-GAN in terms of FID score on CelebA and 3D Chairs dataset.
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。