Title :
An Improved Exponentiated stochastic gradient algorithm
Author :
Rusu, Corneliu ; Cowan, Colin F N
Author_Institution :
Fac. of Electron., Telecom, & Inf. Tech., Tech. Univ. of Cluj-Napoca, Cluj-Napoca, Romania
fDate :
May 30 2010-June 2 2010
Abstract :
Recently, few stochastic gradient algorithms have been proposed and they are based on cost functions that have exponential dependence on the chosen error. However, we have experienced that the proposed cost function based on exponential of the squared error does not converge always. In this paper we modify this cost function in order to assure the convergence and a new IE (Improved Exponentiated) stochastic gradient algorithm has been obtained. The resulting method has attractive properties in both stationary and abrupt-change situations.
Keywords :
FIR filters; adaptive filters; convergence of numerical methods; gradient methods; least mean squares methods; stochastic processes; adaptive FIR filter; cost function; exponential dependence; improved exponentiated stochastic gradient algorithm; least-mean-square; squared error exponential; Adaptive filters; Adaptive signal processing; Convergence; Cost function; Finite impulse response filter; Least squares approximation; Signal processing algorithms; Stochastic processes; Stochastic resonance; Telecommunications;
Conference_Titel :
Circuits and Systems (ISCAS), Proceedings of 2010 IEEE International Symposium on
Conference_Location :
Paris
Print_ISBN :
978-1-4244-5308-5
Electronic_ISBN :
978-1-4244-5309-2
DOI :
10.1109/ISCAS.2010.5538010
