We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: RG-Flow: A hierarchical and explainable flow model based on renormalization group and sparse prior

Abstract: Flow-based generative models have become an important class of unsupervised learning approaches. In this work, we incorporate the key idea of renormalization group (RG) and sparse prior distribution to design a hierarchical flow-based generative model, called RG-Flow, which can separate information at different scales of images with disentangled representations at each scale. We demonstrate our method mainly on the CelebA dataset and show that the disentangled representations at different scales enable semantic manipulation and style mixing of the images. To visualize the latent representations, we introduce receptive fields for flow-based models and find that the receptive fields learned by RG-Flow are similar to those in convolutional neural networks. In addition, we replace the widely adopted Gaussian prior distribution by a sparse prior distribution to further enhance the disentanglement of representations. From a theoretical perspective, the proposed method has $O(\log L)$ complexity for image inpainting compared to previous generative models with $O(L^2)$ complexity.
Comments: 9 pages, 7 figures, with appendix and the newly added multi-scale experiments
Subjects: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (stat.ML)
Cite as: arXiv:2010.00029 [cs.LG]
  (or arXiv:2010.00029v4 [cs.LG] for this version)

Submission history

From: Hong-Ye Hu [view email]
[v1] Wed, 30 Sep 2020 18:04:04 GMT (18742kb,D)
[v2] Fri, 2 Oct 2020 18:27:26 GMT (18898kb,D)
[v3] Wed, 7 Oct 2020 00:59:10 GMT (18905kb,D)
[v4] Fri, 18 Dec 2020 20:27:11 GMT (11165kb,D)

Link back to: arXiv, form interface, contact.