Leveraging sparse and shared feature activations for disentangled representation learning

Fumero M, Wenzel F, Zancato L, Achille A, Rodolà E, Soatto S, Schölkopf B, Locatello F. Leveraging sparse and shared feature activations for disentangled representation learning. arXiv, 2304.07939.

Download (ext.)

Preprint | Submitted | English
Author
Fumero, Marco; Wenzel, Florian; Zancato, Luca; Achille, Alessandro; Rodolà, Emanuele; Soatto, Stefano; Schölkopf, Bernhard; Locatello, FrancescoISTA
Department
Abstract
Recovering the latent factors of variation of high dimensional data has so far focused on simple synthetic settings. Mostly building on unsupervised and weakly-supervised objectives, prior work missed out on the positive implications for representation learning on real world data. In this work, we propose to leverage knowledge extracted from a diversified set of supervised tasks to learn a common disentangled representation. Assuming each supervised task only depends on an unknown subset of the factors of variation, we disentangle the feature space of a supervised multi-task model, with features activating sparsely across different tasks and information being shared as appropriate. Importantly, we never directly observe the factors of variations but establish that access to multiple tasks is sufficient for identifiability under sufficiency and minimality assumptions. We validate our approach on six real world distribution shift benchmarks, and different data modalities (images, text), demonstrating how disentangled representations can be transferred to real settings.
Publishing Year
Date Published
2023-04-17
Journal Title
arXiv
Article Number
2304.07939
IST-REx-ID

Cite this

Fumero M, Wenzel F, Zancato L, et al. Leveraging sparse and shared feature activations for disentangled representation learning. arXiv. doi:10.48550/arXiv.2304.07939
Fumero, M., Wenzel, F., Zancato, L., Achille, A., Rodolà, E., Soatto, S., … Locatello, F. (n.d.). Leveraging sparse and shared feature activations for disentangled representation learning. arXiv. https://doi.org/10.48550/arXiv.2304.07939
Fumero, Marco, Florian Wenzel, Luca Zancato, Alessandro Achille, Emanuele Rodolà, Stefano Soatto, Bernhard Schölkopf, and Francesco Locatello. “Leveraging Sparse and Shared Feature Activations for Disentangled Representation Learning.” ArXiv, n.d. https://doi.org/10.48550/arXiv.2304.07939.
M. Fumero et al., “Leveraging sparse and shared feature activations for disentangled representation learning,” arXiv. .
Fumero M, Wenzel F, Zancato L, Achille A, Rodolà E, Soatto S, Schölkopf B, Locatello F. Leveraging sparse and shared feature activations for disentangled representation learning. arXiv, 2304.07939.
Fumero, Marco, et al. “Leveraging Sparse and Shared Feature Activations for Disentangled Representation Learning.” ArXiv, 2304.07939, doi:10.48550/arXiv.2304.07939.
All files available under the following license(s):
Copyright Statement:
This Item is protected by copyright and/or related rights. [...]

Link(s) to Main File(s)
Access Level
OA Open Access

Export

Marked Publications

Open Data ISTA Research Explorer

Sources

arXiv 2304.07939

Search this title in

Google Scholar