%0 Journal Article %1 su2018fvaes %A Su, Jianlin %A Wu, Guang %D 2018 %K flows generative-models %T f-VAEs: Improve VAEs with Conditional Flows %U http://arxiv.org/abs/1809.05861 %X 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.

@article{su2018fvaes, abstract = {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.}, added-at = {2019-12-09T11:33:35.000+0100}, author = {Su, Jianlin and Wu, Guang}, biburl = {https://www.bibsonomy.org/bibtex/20795b32628181f802fc86ceeef11daa1/kirk86}, description = {[1809.05861] f-VAEs: Improve VAEs with Conditional Flows}, interhash = {a63e01ea67dcd29e9b68180922791443}, intrahash = {0795b32628181f802fc86ceeef11daa1}, keywords = {flows generative-models}, note = {cite arxiv:1809.05861}, timestamp = {2019-12-09T11:33:35.000+0100}, title = {f-VAEs: Improve VAEs with Conditional Flows}, url = {http://arxiv.org/abs/1809.05861}, year = 2018 }