4 Hits in 4.2 sec

SeGMA: Semi-Supervised Gaussian Mixture Auto-Encoder [article]

Marek Śmieja, Maciej Wołczyk, Jacek Tabor, Bernhard C. Geiger
2020 arXiv   pre-print
We propose a semi-supervised generative model, SeGMA, which learns a joint probability distribution of data and their classes and which is implemented in a typical Wasserstein auto-encoder framework.  ...  While SeGMA preserves all properties of its semi-supervised predecessors and achieves at least as good generative performance on standard benchmark data sets, it presents additional features: (a) interpolation  ...  Our model, which we will call semi-supervised Gaussian mixture auto-encoder (SeGMA), is an adaptation of the Cramer-Wold auto-encoder (CWAE) [13] , which is an instance of WAE models with maximum mean  ... 
arXiv:1906.09333v2 fatcat:crk7hs6i2fgb3h4v7os6yc2xue

Variational Information Bottleneck for Semi-Supervised Classification

Slava Voloshynovskiy, Olga Taran, Mouad Kondah, Taras Holotyak, Danilo Rezende
2020 Entropy  
We propose a new formulation of semi-supervised IB with hand crafted and learnable priors and link it to the previous methods such as semi-supervised versions of VAE (M1 + M2), AAE, CatGAN, etc.  ...  In this paper, we consider an information bottleneck (IB) framework for semi-supervised classification with several families of priors on latent space representation.  ...  The latent space of SeGMA is assumed to follow a mixture of Gaussians.  ... 
doi:10.3390/e22090943 pmid:33286710 pmcid:PMC7597214 fatcat:chlv5b45qreknpghsaguyt2dce

Cramer-Wold Auto-Encoder

Szymon Knop, Przemyslaw Spurek, Jacek Tabor, Igor T. Podolak, Marcin Mazur, Stanislaw Jastrzebski
2020 Journal of machine learning research  
Inspired by prior works on the Sliced-Wasserstein Auto-Encoders (SWAE) and the Wasserstein Auto-Encoders with MMD-based penalty (WAE-MMD), we propose a new generative model -a Cramer-Wold Auto-Encoder  ...  Its main distinguishing feature is that it has a closed-form of the kernel product of radial Gaussians.  ...  Śmieja et al. (2019) present a semi-supervised generative model SeGMA, which is able to learn a joint probability distribution of data and their classes.  ... 
dblp:journals/jmlr/KnopSTPMJ20 fatcat:mwhr3k6jlbfdribuxrvl7wfi6u

2021 Index IEEE Transactions on Neural Networks and Learning Systems Vol. 32

2021 IEEE Transactions on Neural Networks and Learning Systems  
., +, TNNLS May 2021 2209-2223 SeGMA: Semi-Supervised Gaussian Mixture Autoencoder.  ...  Tolooshams, B., +, TNNLS June 2021 2415-2429 SeGMA: Semi-Supervised Gaussian Mixture Autoencoder.  ... 
doi:10.1109/tnnls.2021.3134132 fatcat:2e7comcq2fhrziselptjubwjme