Optimizing the parSOM Neural Network Implementation for Data Mining with Distributed Memory Systems and Cluster Computing

Authors:
P. Tomsich;A. Rauber;D. Merkl
Affiliations:
-;-;-
Venue:
DEXA '00 Proceedings of the 11th International Workshop on Database and Expert Systems Applications
Year:
2000

Citing 0
Cited 6

ACODF: a novel data clustering approach for data mining in large databases

Journal of Systems and Software - Special issue: Performance modeling and analysis of computer systems and networks
Design and Implementation of Parallel Counterpropagation Networks Using MPI

Informatica
The investigation of discovering potential musical instruments teachers by effective data clustering scheme

WSEAS Transactions on Computers
Spectral Imaging II: Plant disease detection based on data fusion of hyper-spectral and multi-spectral fluorescence imaging using Kohonen maps

Real-Time Imaging
Gene prediction in metagenomic libraries using the self organising map and high performance computing techniques

GCCB'06 Proceedings of the 2006 international conference on Distributed, high-performance and grid computing in computational biology
Mining massive datasets by an unsupervised parallel clustering on a GRID: Novel algorithms and case study

Future Generation Computer Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

The self-organizing map is a prominent unsupervised neural network model which lends itself to the analysis of high-dimensional input data and data mining applications. However, the high execution times required to train the map limit its application in many high-performance data analysis application domains. We discuss the /sub par/SOM implementation, a software-based parallel implementation of the self-organizing map, and its optimization for the analysis of high-dimensional input data using distributed memory systems and clusters. The original /sub par/SOM algorithm scales very well in a parallel execution environment with low communication latencies and exploits parallelism to cope with memory latencies. However it suffers from poor scalability on distributed memory computers. We present optimizations to further decouple the subprocesses, simplify the communication model and improve the portability of the system.