DocumentCode :
3037998
Title :
Convergence of coefficient regularized fully online algorithm
Author :
Tian, Ming-Dang ; Sheng, Bao-Huai
Author_Institution :
Dept. of Math., Shaoxing Univ., Shaoxing, China
fYear :
2011
fDate :
26-28 July 2011
Firstpage :
2059
Lastpage :
2065
Abstract :
Abstract-This paper gives the convergence of coefficient regularized fully online nonsmooth classification algorithm. With the strongly convex loss function based on the Euclidean Space and the parameter λt changes with learning step give a better convergence rate than the usual convex loss functions.
Keywords :
convex programming; learning (artificial intelligence); pattern classification; Euclidean Space; coefficient regularized fully online algorithm convergence; convex loss function; machine learning method; nonsmooth classification algorithm; Approximation algorithms; Classification algorithms; Convergence; Equations; Hilbert space; Kernel; Machine learning algorithms; binary classification; convergence analysis; learning rates; online algorithm;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Multimedia Technology (ICMT), 2011 International Conference on
Conference_Location :
Hangzhou
Print_ISBN :
978-1-61284-771-9
Type :
conf
DOI :
10.1109/ICMT.2011.6002468
Filename :
6002468
Link To Document :
بازگشت