Title :
Low-storage, second-order learning algorithms: an extended abstract
Author :
Schneider, M.H. ; Farrow, Kristan ; Neti, C.
Author_Institution :
Dept. of Math. Sci., Johns Hopkins Univ., Baltimore, MD
Abstract :
Summary form only given, as follows. Variants of Newton and quasi-Newton methods are discussed that require neither construction of matrices nor solution of systems of equations. These methods were applied to solve estimation problems arising in feedforward neural network models. Numerical results were obtained for sonar classification problems
Keywords :
approximation theory; learning systems; neural nets; pattern recognition; sonar; estimation problems; feedforward neural network models; low-storage second-order learning algorithms; quasi-Newton methods; sonar classification; Equations; Feedforward neural networks; Feedforward systems; Neural networks; Sonar;
Conference_Titel :
Neural Networks, 1991., IJCNN-91-Seattle International Joint Conference on
Conference_Location :
Seattle, WA
Print_ISBN :
0-7803-0164-1
DOI :
10.1109/IJCNN.1991.155592
