Applications of approximate inference and experimental design for sparse (generalised) linear models

1 hour 9 mins 11 secs, 63.34 MB, MP3 44100 Hz, 125.0 kbits/sec

Share this media item:

Embed this media item:

<iframe width="_width_" height="_height_" src="https://sms.cam.ac.uk/media/1161/embed" frameborder="0" scrolling="no" allowfullscreen></iframe>

Choose size:

About this item

Available Formats

About this item

Description:	Seeger, MW (MPI for Biological Cybernetics) Friday 27 June 2008, 11:30-12:30 Future Directions in High-Dimensional Data Analysis

Created:

2008-07-15 09:53

Collection:

Statistical Theory and Methods for Complex, High-Dimensional Data

Publisher:

Isaac Newton Institute

Seeger, MW

Language:

eng (English)

Credits:

Author:

Seeger, MW


Abstract:	Sparsity, or more general sub-Gaussianity, is a fundamental regularization principle for high-dimensional statistics. A recent surge of activity has clarified the behaviour of efficient sparse estimators in the worst case, but much less is known about practically efficient approximations to Bayesian inference, which is required for higher-level tasks such as experimental design. We present an efficient framework for Bayesian inference on generalized linear models with sparsity priors, based on the expectation propagation algorithm, a deterministic variational approximation. We highlight some applications where this framework produces promising results. We hope to convey the relevance of approximate inference methods in practice, which substantially go beyond point estimation, yet whose theoretical properties and algorithmic scalability remains insufficiently understood.

Abstract:

Sparsity, or more general sub-Gaussianity, is a fundamental regularization principle for high-dimensional statistics. A recent surge of activity has clarified the behaviour of efficient sparse estimators in the worst case, but much less is known about practically efficient approximations to Bayesian inference, which is required for higher-level tasks such as experimental design.

We present an efficient framework for Bayesian inference on generalized linear models with sparsity priors, based on the expectation propagation algorithm, a deterministic variational approximation. We highlight some applications where this framework produces promising results. We hope to convey the relevance of approximate inference methods in practice, which substantially go beyond point estimation, yet whose theoretical properties and algorithmic scalability remains insufficiently understood.

Available Formats

Format	Quality	Bitrate	Size
MPEG-4 Video	480x360	1.84 Mbits/sec	958.80 MB	View	Download
WebM	480x360	678.21 kbits/sec	342.75 MB	View	Download
Flash Video	480x360	806.21 kbits/sec	409.90 MB	View	Download
iPod Video	480x360	505.25 kbits/sec	256.88 MB	View	Download
QuickTime	384x288	848.84 kbits/sec	431.57 MB	View	Download
MP3 *	44100 Hz	125.0 kbits/sec	63.34 MB	Listen	Download
Auto	(Allows browser to choose a format it supports)