Linear Independent Component Analysis via Optimal Transport
Abstract
Linear Independent Component Analysis (ICA) recovers jointly independent source signals from their linear mixtures. To achieve this, classical ICA algorithms attempt to maximize non-Gaussianity, measured by negentropy, which is linked to independence by information theory. Because exact negentropy optimization is intractable, they rely on proxy contrast functions, such as fourth-order cumulants and parametric log-likelihoods. We propose instead to use the squared L2-Wasserstein distance to a standard Gaussian as the ICA contrast. We show that the Wasserstein distance between a standard normal distribution and linear projections of the data is maximized when the projection recovers an independent component, and that under a regularity condition on the sources this maximum is separated from every genuine mixture by an explicit margin. We uncover the advantageous properties of the resulting estimator: for sources with a smooth density it is √N-consistent and asymptotically normal with a closed-form variance, whereas for a source with an atom the contrast has a kink at the true direction, which makes the estimator exact with probability tending to one. The proposed OT-ICA algorithm finds this projection by gradient-based optimization. Empirical evaluation on simulated data shows that OT-ICA outperforms proxy-based methods and that the W22 contrast provides more robust signal than proxy contrasts in measuring non-Gaussianity for different distributional mixtures of the latent variables. Application to linear causal disentanglement and EEG artifact removal, along with further applications detailed in the appendix, confirms OT-ICA can be used for applied ICA tasks without distributional assumptions.
Related
- On the Linear Convergence of SISTA for Entropic Martingale Optimal Transport
- Annealed Sinkhorn with Momentum: Certified Unregularized Optimal Transport in Linear Memory
- Globally Convergent Block Coordinate Descent Algorithms for Semi-Nonnegative Independent Component Analysis
- Identifiability from Moments in Independent Component Analysis Mixture Models
- Principal component analysis in Φ α-linear spaces
BibTeX
@misc{jha2026linear,
title = {{Linear Independent Component Analysis via Optimal Transport}},
author = {Ashutosh Jha and Michel Besserve and Simon Buchholz},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2607.14081},
eprint = {2607.14081},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.14081},
}
RIS
TY - UNPB
TI - Linear Independent Component Analysis via Optimal Transport
AU - Ashutosh Jha
AU - Michel Besserve
AU - Simon Buchholz
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2607.14081
UR - https://arxiv.org/abs/2607.14081
AB - Linear Independent Component Analysis (ICA) recovers jointly independent source signals from their linear mixtures. To achieve this, classical ICA algorithms attempt to maximize non-Gaussianity, measured by negentropy, which is linked to independence by information theory. Because exact negentropy optimization is intractable, they rely on proxy contrast functions, such as fourth-order cumulants and parametric log-likelihoods. We propose instead to use the squared L_2-Wasserstein distance to a standard Gaussian as the ICA contrast. We show that the Wasserstein distance between a standard normal distribution and linear projections of the data is maximized when the projection recovers an independent component, and that under a regularity condition on the sources this maximum is separated from every genuine mixture by an explicit margin. We uncover the advantageous properties of the resulting estimator: for sources with a smooth density it is √N-consistent and asymptotically normal with a closed-form variance, whereas for a source with an atom the contrast has a kink at the true direction, which makes the estimator exact with probability tending to one. The proposed OT-ICA algorithm finds this projection by gradient-based optimization. Empirical evaluation on simulated data shows that OT-ICA outperforms proxy-based methods and that the W_2^2 contrast provides more robust signal than proxy contrasts in measuring non-Gaussianity for different distributional mixtures of the latent variables. Application to linear causal disentanglement and EEG artifact removal, along with further applications detailed in the appendix, confirms OT-ICA can be used for applied ICA tasks without distributional assumptions.
ER -