2022-06-28-lachapelle22a.md

abstract	booktitle	title	year	layout	series	publisher	issn	id	month	tex_title	firstpage	lastpage	page	order	cycles	bibtex_author	author	date	address	container-title	volume	genre	issued	pdf	extras
This work introduces a novel principle we call disentanglement via mechanism sparsity regularization, which can be applied when the latent factors of interest depend sparsely on past latent factors and/or observed auxiliary variables. We propose a representation learning method that induces disentanglement by simultaneously learning the latent factors and the sparse causal graphical model that relates them. We develop a rigorous identifiability theory, building on recent nonlinear independent component analysis (ICA) results, that formalizes this principle and shows how the latent variables can be recovered up to permutation if one regularizes the latent mechanisms to be sparse and if some graph connectivity criterion is satisfied by the data generating process. As a special case of our framework, we show how one can leverage unknown-target interventions on the latent factors to disentangle them, thereby drawing further connections between ICA and causality. We propose a VAE-based method in which the latent mechanisms are learned and regularized via binary masks, and validate our theory by showing it learns disentangled representations in simulations.	First Conference on Causal Learning and Reasoning	Disentanglement via Mechanism Sparsity Regularization: A New Principle for Nonlinear ICA	2022	inproceedings	Proceedings of Machine Learning Research	PMLR	2640-3498	lachapelle22a	0	Disentanglement via Mechanism Sparsity Regularization: A New Principle for Nonlinear {ICA}	428	484	428-484	428	false	Lachapelle, Sebastien and Rodriguez, Pau and Sharma, Yash and Everett, Katie E and PRIOL, R{\'e}mi LE and Lacoste, Alexandre and Lacoste-Julien, Simon	given family Sebastien Lachapelle given family Pau Rodriguez given family Yash Sharma given family Katie E Everett given family Rémi LE PRIOL given family Alexandre Lacoste given family Simon Lacoste-Julien	2022-06-28		Proceedings of the First Conference on Causal Learning and Reasoning	177	inproceedings	date-parts 2022 6 28	https://proceedings.mlr.press/v177/lachapelle22a/lachapelle22a.pdf