论文信息 - Variational Bayesian Independent Component Analysis

Variational Bayesian Independent Component Analysis

Blind separation of signals through the info-max algorithm may be viewed as maximum likelihood learning in a latent variable m odel. In this paper we present an alternative approach to maximum lik elihood learning in these models, namely Bayesian inference. It has already been shown how Bayesian inference can be applied to determine lat ent dimensionality in principal component analysis models (Bishop, 1999a). In this paper we derive a similar approach for removing unecess ary source dimensions in an independent component analysis model. We p resent results on a toy data-set and on some artificially mixed image s.

Neil D. Lawrence | Christopher M. Bishop | Charles M. Bishop | Neil D. Lawrence

[1] Michael I. Jordan,et al. An Introduction to Variational Methods for Graphical Models , 1999, Machine Learning.

[2] David Mackay,et al. Probable networks and plausible predictions - a review of practical Bayesian methods for supervised neural networks , 1995 .

[3] Terrence J. Sejnowski,et al. An Information-Maximization Approach to Blind Separation and Blind Deconvolution , 1995, Neural Computation.

[4] Hagai Attias,et al. Independent Factor Analysis , 1999, Neural Computation.

[5] J. Cardoso. Infomax and maximum likelihood for blind source separation , 1997, IEEE Signal Processing Letters.

[6] D. Chakrabarti,et al. A fast fixed - point algorithm for independent component analysis , 1997 .

[7] David J. C. MacKay,et al. Developments in Probabilistic Modelling with Neural Networks - Ensemble Learning , 1995, SNN Symposium on Neural Networks.

[8] Harri Lappalainen,et al. Ensemble learning for independent component analysis , 1999 .

[9] Neil D. Lawrence,et al. A Variational B ayesian Committee of Neural Networks , 1999 .

[10] Charles M. Bishop. Variational principal components , 1999 .