SPLICE: Fully Tractable Hierarchical Extension of ICA with Pooling

Jun-ichiro Hirayama, Aapo Hyvärinen, Motoaki Kawanabe
Proceedings of the 34th International Conference on Machine Learning, PMLR 70:1491-1500, 2017.

Abstract

We present a novel probabilistic framework for a hierarchical extension of independent component analysis (ICA), with a particular motivation in neuroscientific data analysis and modeling. The framework incorporates a general subspace pooling with linear ICA-like layers stacked recursively. Unlike related previous models, our generative model is fully tractable: both the likelihood and the posterior estimates of latent variables can readily be computed with analytically simple formulae. The model is particularly simple in the case of complex-valued data since the pooling can be reduced to taking the modulus of complex numbers. Experiments on electroencephalography (EEG) and natural images demonstrate the validity of the method.

Cite this Paper

BibTeX
@InProceedings{pmlr-v70-hirayama17a, title = {{SPLICE}: Fully Tractable Hierarchical Extension of {ICA} with Pooling}, author = {Jun-ichiro Hirayama and Aapo Hyv{\"a}rinen and Motoaki Kawanabe}, booktitle = {Proceedings of the 34th International Conference on Machine Learning}, pages = {1491--1500}, year = {2017}, editor = {Precup, Doina and Teh, Yee Whye}, volume = {70}, series = {Proceedings of Machine Learning Research}, month = {06--11 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v70/hirayama17a/hirayama17a.pdf}, url = {https://proceedings.mlr.press/v70/hirayama17a.html}, abstract = {We present a novel probabilistic framework for a hierarchical extension of independent component analysis (ICA), with a particular motivation in neuroscientific data analysis and modeling. The framework incorporates a general subspace pooling with linear ICA-like layers stacked recursively. Unlike related previous models, our generative model is fully tractable: both the likelihood and the posterior estimates of latent variables can readily be computed with analytically simple formulae. The model is particularly simple in the case of complex-valued data since the pooling can be reduced to taking the modulus of complex numbers. Experiments on electroencephalography (EEG) and natural images demonstrate the validity of the method.} }
Endnote
%0 Conference Paper %T SPLICE: Fully Tractable Hierarchical Extension of ICA with Pooling %A Jun-ichiro Hirayama %A Aapo Hyvärinen %A Motoaki Kawanabe %B Proceedings of the 34th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2017 %E Doina Precup %E Yee Whye Teh %F pmlr-v70-hirayama17a %I PMLR %P 1491--1500 %U https://proceedings.mlr.press/v70/hirayama17a.html %V 70 %X We present a novel probabilistic framework for a hierarchical extension of independent component analysis (ICA), with a particular motivation in neuroscientific data analysis and modeling. The framework incorporates a general subspace pooling with linear ICA-like layers stacked recursively. Unlike related previous models, our generative model is fully tractable: both the likelihood and the posterior estimates of latent variables can readily be computed with analytically simple formulae. The model is particularly simple in the case of complex-valued data since the pooling can be reduced to taking the modulus of complex numbers. Experiments on electroencephalography (EEG) and natural images demonstrate the validity of the method.
APA
Hirayama, J., Hyvärinen, A. & Kawanabe, M.. (2017). SPLICE: Fully Tractable Hierarchical Extension of ICA with Pooling. Proceedings of the 34th International Conference on Machine Learning, in Proceedings of Machine Learning Research 70:1491-1500 Available from https://proceedings.mlr.press/v70/hirayama17a.html.

Related Material