Latent Bernoulli Autoencoder

Jiri Fajtl, Vasileios Argyriou, Dorothy Monekosso, Paolo Remagnino
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:2964-2974, 2020.

Abstract

In this work, we pose the question whether it is possible to design and train an autoencoder model in an end-to-end fashion to learn representations in the multivariate Bernoulli latent space, and achieve performance comparable with the state-of-the-art variational methods. Moreover, we investigate how to generate novel samples and perform smooth interpolation and attributes modification in the binary latent space. To meet our objective, we propose a simplified, deterministic model with a straight-through gradient estimator to learn the binary latents and show its competitiveness with the latest VAE methods. Furthermore, we propose a novel method based on a random hyperplane rounding for sampling and smooth interpolation in the latent space. Our method performs on a par or better than the current state-of-the-art methods on common CelebA, CIFAR-10 and MNIST datasets.

Cite this Paper


BibTeX
@InProceedings{pmlr-v119-fajtl20a, title = {Latent Bernoulli Autoencoder}, author = {Fajtl, Jiri and Argyriou, Vasileios and Monekosso, Dorothy and Remagnino, Paolo}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {2964--2974}, year = {2020}, editor = {Hal Daumé III and Aarti Singh}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/fajtl20a/fajtl20a.pdf}, url = { http://proceedings.mlr.press/v119/fajtl20a.html }, abstract = {In this work, we pose the question whether it is possible to design and train an autoencoder model in an end-to-end fashion to learn representations in the multivariate Bernoulli latent space, and achieve performance comparable with the state-of-the-art variational methods. Moreover, we investigate how to generate novel samples and perform smooth interpolation and attributes modification in the binary latent space. To meet our objective, we propose a simplified, deterministic model with a straight-through gradient estimator to learn the binary latents and show its competitiveness with the latest VAE methods. Furthermore, we propose a novel method based on a random hyperplane rounding for sampling and smooth interpolation in the latent space. Our method performs on a par or better than the current state-of-the-art methods on common CelebA, CIFAR-10 and MNIST datasets.} }
Endnote
%0 Conference Paper %T Latent Bernoulli Autoencoder %A Jiri Fajtl %A Vasileios Argyriou %A Dorothy Monekosso %A Paolo Remagnino %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-fajtl20a %I PMLR %P 2964--2974 %U http://proceedings.mlr.press/v119/fajtl20a.html %V 119 %X In this work, we pose the question whether it is possible to design and train an autoencoder model in an end-to-end fashion to learn representations in the multivariate Bernoulli latent space, and achieve performance comparable with the state-of-the-art variational methods. Moreover, we investigate how to generate novel samples and perform smooth interpolation and attributes modification in the binary latent space. To meet our objective, we propose a simplified, deterministic model with a straight-through gradient estimator to learn the binary latents and show its competitiveness with the latest VAE methods. Furthermore, we propose a novel method based on a random hyperplane rounding for sampling and smooth interpolation in the latent space. Our method performs on a par or better than the current state-of-the-art methods on common CelebA, CIFAR-10 and MNIST datasets.
APA
Fajtl, J., Argyriou, V., Monekosso, D. & Remagnino, P.. (2020). Latent Bernoulli Autoencoder. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:2964-2974 Available from http://proceedings.mlr.press/v119/fajtl20a.html .

Related Material