%0 Generic %1 voynov2020unsupervised %A Voynov, Andrey %A Babenko, Artem %D 2020 %K cv dl gan %T Unsupervised Discovery of Interpretable Directions in the GAN Latent Space %U http://arxiv.org/abs/2002.03754 %X The latent spaces of typical GAN models often have semantically meaningful directions. Moving in these directions corresponds to human-interpretable image transformations, such as zooming or recoloring, enabling a more controllable generation process. However, the discovery of such directions is currently performed in a supervised manner, requiring human labels, pretrained models, or some form of self-supervision. These requirements can severely limit a range of directions existing approaches can discover. In this paper, we introduce an unsupervised method to identify interpretable directions in the latent space of a pretrained GAN model. By a simple model-agnostic procedure, we find directions corresponding to sensible semantic manipulations without any form of (self-)supervision. Furthermore, we reveal several non-trivial findings, which would be difficult to obtain by existing methods, e.g., a direction corresponding to background removal. As an immediate practical benefit of our work, we show how to exploit this finding to achieve a new state-of-the-art for the problem of saliency detection.

@misc{voynov2020unsupervised, abstract = {The latent spaces of typical GAN models often have semantically meaningful directions. Moving in these directions corresponds to human-interpretable image transformations, such as zooming or recoloring, enabling a more controllable generation process. However, the discovery of such directions is currently performed in a supervised manner, requiring human labels, pretrained models, or some form of self-supervision. These requirements can severely limit a range of directions existing approaches can discover. In this paper, we introduce an unsupervised method to identify interpretable directions in the latent space of a pretrained GAN model. By a simple model-agnostic procedure, we find directions corresponding to sensible semantic manipulations without any form of (self-)supervision. Furthermore, we reveal several non-trivial findings, which would be difficult to obtain by existing methods, e.g., a direction corresponding to background removal. As an immediate practical benefit of our work, we show how to exploit this finding to achieve a new state-of-the-art for the problem of saliency detection.}, added-at = {2020-02-14T20:24:32.000+0100}, author = {Voynov, Andrey and Babenko, Artem}, biburl = {https://www.bibsonomy.org/bibtex/2f974b2969d31e2b1b03c2f633e372eda/bechr7}, description = {[2002.03754] Unsupervised Discovery of Interpretable Directions in the GAN Latent Space}, interhash = {2a08222eac5ea5db6533b526a9382bbe}, intrahash = {f974b2969d31e2b1b03c2f633e372eda}, keywords = {cv dl gan}, note = {cite arxiv:2002.03754}, timestamp = {2020-02-14T20:24:32.000+0100}, title = {Unsupervised Discovery of Interpretable Directions in the GAN Latent Space}, url = {http://arxiv.org/abs/2002.03754}, year = 2020 }