Li, X., Lin, C. orcid.org/0000-0003-3454-2468, Li, R. et al. (2 more authors) (2020) Latent space factorisation and manipulation via matrix subspace projection. In: Daumé, H. and Singh, A., (eds.) Proceedings of the International Conference on Machine Learning 1 pre-proceedings (ICML 2020). The 37th International Conference on Machine Learning (ICML), 13-18 Jul 2020, Online conference. ICML
Abstract
We tackle the problem disentangling the latent space of an autoencoder in order to separate labelled attribute information from other characteristic information. This then allows us to change selected attributes while preserving other information. Our method, matrix subspace projection, is much simpler than previous approaches to latent space factorisation, for example not requiring multiple discriminators or a careful weighting among loss functions. Furthermore our new model can be applied to autoencoders as a plugin, and works across diverse domains such as images or text. We demonstrate the utility of our method for attribute manipulation in autoencoders trained across varied domains, using both human evaluation and automated methods. The quality of generation of our new model (e.g. reconstruction, conditional generation) is highly competitive to a number of strong baselines.
Metadata
Item Type: | Proceedings Paper |
---|---|
Authors/Creators: |
|
Editors: |
|
Copyright, Publisher and Additional Information: | © 2020 The Authors. |
Dates: |
|
Institution: | The University of Sheffield |
Academic Units: | The University of Sheffield > Faculty of Engineering (Sheffield) > Department of Computer Science (Sheffield) |
Depositing User: | Symplectic Sheffield |
Date Deposited: | 03 Nov 2020 08:44 |
Last Modified: | 13 Aug 2021 07:39 |
Published Version: | https://proceedings.icml.cc/paper/2020/hash/61b1fb... |
Status: | Published online |
Publisher: | ICML |
Refereed: | Yes |
Related URLs: | |
Open Archives Initiative ID (OAI ID): | oai:eprints.whiterose.ac.uk:167435 |