BSBL: Block Sparse Bayesian Learning

I developed fast implementations of the Block Sparse Bayesian Learning (BSBL, see Zhilin Zhang's Website) Framework, using the Fast Marginalized Likelihood Maximization method (see Fast marginal likelihood maximisation for sparse Beysian models).

BSBL-FM: A Fast Marginalized Implementation

see the github repository for more details : bsbl-fm @ github

Basic

The BSBL-FM is a fast implementation of the BSBL framework. It is based on BCS (Ji2008) and fastRVM (Tipping2003). The major contributions are the block extension and the exploiting of intra-block correlation. The BSBL-FM algorithm is extremely fast when the signal has block structure and low block sparsity level.

Demos

Papers

1. Benyuan Liu, Zhilin Zhang, Gary Xu, Hongqi Fan, Qiang Fu, Energy Efficient Telemonitoring of Physiological Signals via Compressed Sensing: A Fast Algorithm and Power Consumption Evaluation. Biomedical Signal Processing and Control (BSPC), 2014 Vol 11, pp 80–88. Arxiv Preprint, 1309.7843.
   DOI:10.1016/j.bspc.2014.02.010

2. Benyuan Liu, Zhilin Zhang, Hongqi Fan, Qiang Fu, Compression via Compressive Sensing: A Low-Power Framework for the Telemonitoring of Multi-Channel Physiological Signals. International Workshop on Biomedical and Health Informatics (BHI2013, in conjunction with IEEE International Conference on Bioinformatics and Biomedicine BIBM), Shanghai, China, Dec. 2013, pp 9-12. Arxiv Preprint, 1309.4136.
   DOI:10.1109/BIBM.2013.6732592

3. Benyuan Liu, Zhilin Zhang, Hongqi Fan, Fast Marginalized Block Sparse Bayesian Learning Algorithm. Arxiv Preprint, 1211.4909.

Implementations

1. Matlab Version : The Matlab Implementation can be downloaded at: bsbl-fm @ github.

2. C Version : The C version as well as the GPU Cuda Implementation will be available soon.

3. Python Version : The Python implementations are available at: pyBSBL @ github.