Software for the Future
Posted: 7 Jan 2015 | 10:18
Big-Data compressive sensing: fast, parallelised and distributed algorithms
EPCC is excited to be part of a new project, funded through EPSRC's Software for the Future programme, to develop and exploit compressive-sensing algorithms for large-scale data problems.
A team of astrophysicists, engineers and computer scientists from Edinburgh University, Heriot Watt University, and University College London, will lead research into imaging techniques that exploit the mathematical theory of compressive sampling to recover complex natural images from partial information. The theory provides both new concepts to optimise the data acquisition procedures for radio astronomy and for medical imaging.
Particularly, the project will apply the theory of compressive sampling to significantly accelerate diffusion MRI scans. This would open the door to the clinical application of high-angular-resolution diffusion MRI, which probes the direction of water diffusion at each point in the brain with the aim of reconstructing global neuronal pathways as computer models.
Furthermore, it will investigate ways to redesign radio telescope array configurations, so that each data point contains the maximum amount of information, while novel recovery algorithms will be applied to reconstruct ultra-high resolution images of the sky. These techniques will be important for the upcoming flagship Square Kilometre Array (SKA) telescope which will be constructed in the next 10 years and will provide unprecedent resolution and sensitivity of radio telescope data collection.
EPCC is working with the scientists who are developing the compressive-sensing algorithms that will produce parallelised and optimised solutions for compressive sensing. These will scale to the very large data sets that will be generated by future scientific instruments.
Images: (Top) Reconstructed neuronal connections in the brain. Credit: Thomas Schultz, CC BY-SA
(Above) Close-up artist rendition of the Australian SKA LFAA (Low Frequency Aperture Array) instrument. Credit SKA Organisation.