Blumensath, T. and Davies, M.E.
A fast importance sampling algorithm for unsupervised learning of over-complete dictionaries
At IEEE International Conference on Acoustics, Speech and Signal Processing, United States.
18 - 23 Mar 2005.
Full text not available from this repository.
We use Bayesian statistics to study the dictionary learning problem
in which an over-complete generative signal model has to be
adapted for optimally sparse signal representations. With such a
formulation we develop a stochastic gradient learning algorithm
based on Importance Sampling techniques to minimise the negative
marginal log-likelihood. As this likelihood is not available
analytically, approximations have to be utilised. The Importance
Sampling Monte Carlo marginalisation proposed here improves on
previous methods and addresses three main issues: 1) bias of the
gradient estimate; 2) multi-modality of the distribution to be approximated;
and 3) computational efficiency. Experimental results
show the advantages of the new method when compared to previous
techniques. The gained efficiency allows the treatment of large
scale problems in a statistically sound framework as demonstrated
here by the extraction of individual piano notes from a polyphonic
Conference or Workshop Item
|Venue - Dates:
||IEEE International Conference on Acoustics, Speech and Signal Processing, United States, 2005-03-18 - 2005-03-23
||Signal Processing & Control Grp
||01 Jul 2010 10:42
||18 Apr 2017 04:20
|Further Information:||Google Scholar|
Actions (login required)