Multilinear Latent Conditioning for Generating Unseen Attribute Combinations

Markos Georgopoulos, Grigorios Chrysos, Maja Pantic, Yannis Panagakis
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:3442-3451, 2020.

Abstract

Deep generative models rely on their inductive bias to facilitate generalization, especially for problems with high dimensional data, like images. However, empirical studies have shown that variational autoencoders (VAE) and generative adversarial networks (GAN) lack the generalization ability that occurs naturally in human perception. For example, humans can visualize a woman smiling after only seeing a smiling man. On the contrary, the standard conditional VAE (cVAE) is unable to generate unseen attribute combinations. To this end, we extend cVAE by introducing a multilinear latent conditioning framework that captures the multiplicative interactions between the attributes. We implement two variants of our model and demonstrate their efficacy on MNIST, Fashion-MNIST and CelebA. Altogether, we design a novel conditioning framework that can be used with any architecture to synthesize unseen attribute combinations.

Cite this Paper

BibTeX
@InProceedings{pmlr-v119-georgopoulos20a, title = {Multilinear Latent Conditioning for Generating Unseen Attribute Combinations}, author = {Georgopoulos, Markos and Chrysos, Grigorios and Pantic, Maja and Panagakis, Yannis}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {3442--3451}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/georgopoulos20a/georgopoulos20a.pdf}, url = {https://proceedings.mlr.press/v119/georgopoulos20a.html}, abstract = {Deep generative models rely on their inductive bias to facilitate generalization, especially for problems with high dimensional data, like images. However, empirical studies have shown that variational autoencoders (VAE) and generative adversarial networks (GAN) lack the generalization ability that occurs naturally in human perception. For example, humans can visualize a woman smiling after only seeing a smiling man. On the contrary, the standard conditional VAE (cVAE) is unable to generate unseen attribute combinations. To this end, we extend cVAE by introducing a multilinear latent conditioning framework that captures the multiplicative interactions between the attributes. We implement two variants of our model and demonstrate their efficacy on MNIST, Fashion-MNIST and CelebA. Altogether, we design a novel conditioning framework that can be used with any architecture to synthesize unseen attribute combinations.} }
Endnote
%0 Conference Paper %T Multilinear Latent Conditioning for Generating Unseen Attribute Combinations %A Markos Georgopoulos %A Grigorios Chrysos %A Maja Pantic %A Yannis Panagakis %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-georgopoulos20a %I PMLR %P 3442--3451 %U https://proceedings.mlr.press/v119/georgopoulos20a.html %V 119 %X Deep generative models rely on their inductive bias to facilitate generalization, especially for problems with high dimensional data, like images. However, empirical studies have shown that variational autoencoders (VAE) and generative adversarial networks (GAN) lack the generalization ability that occurs naturally in human perception. For example, humans can visualize a woman smiling after only seeing a smiling man. On the contrary, the standard conditional VAE (cVAE) is unable to generate unseen attribute combinations. To this end, we extend cVAE by introducing a multilinear latent conditioning framework that captures the multiplicative interactions between the attributes. We implement two variants of our model and demonstrate their efficacy on MNIST, Fashion-MNIST and CelebA. Altogether, we design a novel conditioning framework that can be used with any architecture to synthesize unseen attribute combinations.
APA
Georgopoulos, M., Chrysos, G., Pantic, M. & Panagakis, Y.. (2020). Multilinear Latent Conditioning for Generating Unseen Attribute Combinations. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:3442-3451 Available from https://proceedings.mlr.press/v119/georgopoulos20a.html.

Related Material