f-VAEs: Improve VAEs with Conditional Flows
arXiv:1809.05861v121 citations
Originality Incremental advance
AI Analysis
This work addresses image quality and efficiency issues in generative models for machine learning applications, representing an incremental improvement.
The paper tackles the problem of generating blurred images in VAEs and the computational inefficiency of flow-based models by integrating them into f-VAEs, resulting in more vivid images and faster convergence with smaller architectures.
In this paper, we integrate VAEs and flow-based generative models successfully and get f-VAEs. Compared with VAEs, f-VAEs generate more vivid images, solved the blurred-image problem of VAEs. Compared with flow-based models such as Glow, f-VAE is more lightweight and converges faster, achieving the same performance under smaller-size architecture.