Title :
Sub gradient iterative method for neural networks training
Author :
Shcherbakova, Galyna ; Krylov, Victor ; Abakumov, Valentin ; Brovkov, Volodymyr ; Kozina, Iuliia
Author_Institution :
Odessa Nat. Polytech. Univ., Odessa, Ukraine
Abstract :
Sub gradient iterative method in the wavelet transformed domain for neural networks training rule is proposed. This method allows to improve noise stability, to reduce local extreme and initial point search sensitiveness.
Keywords :
gradient methods; learning (artificial intelligence); neural nets; optimisation; search problems; wavelet transforms; initial point search sensitivity; local extreme sensitivity; neural network training rule; noise stability; subgradient iterative method; wavelet transformed domain; Estimation; Neural networks; Neurons; Noise; Optimization; Training; Wavelet transforms; neural networks; sub gradient; wavelet transforming;
Conference_Titel :
Intelligent Data Acquisition and Advanced Computing Systems (IDAACS), 2011 IEEE 6th International Conference on
Conference_Location :
Prague
Print_ISBN :
978-1-4577-1426-9
DOI :
10.1109/IDAACS.2011.6072774
