TY - GEN
T1 - GPU-based biclustering for neural information processing
AU - Lo, Alan W. Y.
AU - Liu, Benben
AU - Cheung, Ray C. C.
PY - 2012
Y1 - 2012
N2 - This paper presents an efficient mapping of geometric biclustering (GBC) algorithm for neural information processing on Graphical Processing Unit (GPU). The proposed designs consist of five different versions which extensively study the use of memory components on the GPU board for mapping the GBC algorithm. GBC algorithm is used to find any maximal biclusters, which are common patterns in each column in the neural processing and gene microarray data. A microarray commonly involves a huge number of data, such as thousands of rows by thousands of columns so that finding the maximal biclusters involves intensive computation. The advantage of GPU is its ability of parallel computing which means that for those independent procedures, they can be carried out at the same time. Experimental results show that the GPU-based GBC could reduce the processing time largely due to the parallel computing of GPU, and its scalability. As an example, GBC algorithm involves a large number of AND operations which utilize the parallel GPU computations, that can be further practically used for other neural processing algorithms. © 2012 Springer-Verlag.
AB - This paper presents an efficient mapping of geometric biclustering (GBC) algorithm for neural information processing on Graphical Processing Unit (GPU). The proposed designs consist of five different versions which extensively study the use of memory components on the GPU board for mapping the GBC algorithm. GBC algorithm is used to find any maximal biclusters, which are common patterns in each column in the neural processing and gene microarray data. A microarray commonly involves a huge number of data, such as thousands of rows by thousands of columns so that finding the maximal biclusters involves intensive computation. The advantage of GPU is its ability of parallel computing which means that for those independent procedures, they can be carried out at the same time. Experimental results show that the GPU-based GBC could reduce the processing time largely due to the parallel computing of GPU, and its scalability. As an example, GBC algorithm involves a large number of AND operations which utilize the parallel GPU computations, that can be further practically used for other neural processing algorithms. © 2012 Springer-Verlag.
KW - Biclustering
KW - Graphics Processing Unit (GPU)
KW - High Performance Computing (HPC)
UR - http://www.scopus.com/inward/record.url?scp=84869015282&partnerID=8YFLogxK
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-84869015282&origin=recordpage
U2 - 10.1007/978-3-642-34500-5_17
DO - 10.1007/978-3-642-34500-5_17
M3 - RGC 32 - Refereed conference paper (with host publication)
SN - 9783642344992
VL - 7667 LNCS
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 134
EP - 141
BT - Neural Information Processing
PB - Springer Verlag
T2 - 19th International Conference on Neural Information Processing (ICONIP 2012)
Y2 - 12 November 2012 through 15 November 2012
ER -