Title :
Adaptive neural network algorithm for solving linear algebra problems
Author :
Galushkin, A.I. ; Sudarikov, V.A.
Author_Institution :
Sci. Neurocomput. Centre, Acad. of Sci., Moscow, Russia
Abstract :
Discusses the construction of adaptive neural network algorithms for solving systems of linear equations and linear inequalities. The performance and efficiency are evaluated for parallel algorithms. An assessment is made of the gain in terms of throughput when neural network algorithms are used in terms of `natural´ implementation. The speed performance of an algorithm is proportional to the number of physically implemented linear threshold elements and may reach 2K
Keywords :
computational complexity; linear algebra; mathematics computing; neural nets; parallel algorithms; adaptive neural network algorithms; efficiency; gain; linear algebra; linear equations; linear inequalities; linear threshold elements; parallel algorithms; performance; throughput; Adaptive systems; Appraisal; Concrete; Equations; Linear algebra; Linear matrix inequalities; Neural networks; Optimization methods; Vectors;
Conference_Titel :
Neuroinformatics and Neurocomputers, 1992., RNNS/IEEE Symposium on
Conference_Location :
Rostov-on-Don
Print_ISBN :
0-7803-0809-3
DOI :
10.1109/RNNS.1992.268602
