DocumentCode :
2392830
Title :
Bayesian online multi-task learning using regularization networks
Author :
Pillonetto, Gianluigi ; Dinuzzo, Francesco ; De Nicolao, G.
Author_Institution :
Dipt. di Ing. dell´´Inf., Padova Univ., Padova
fYear :
2008
fDate :
11-13 June 2008
Firstpage :
4517
Lastpage :
4522
Abstract :
Recently, standard single-task kernel methods have been extended to the case of multi-task learning under the framework of regularization. Experimental results have shown that such an approach can perform much better than single-task techniques, especially when few examples per task are available. However, a possible drawback may be computational complexity. For instance, when using regularization networks, complexity scales as the cube of the overall number of data associated with all the tasks. In this paper, an efficient computational scheme is derived for a widely applied class of multi-task kernels. More precisely, a quadratic loss is assumed and the multi-task kernel is the sum of a common term and a task-specific one. The proposed algorithm performs online learning recursively updating the estimates as new data become available. The learning problem is formulated in a Bayesian setting. The optimal estimates are obtained by solving a sequence of subproblems which involve projection of random variables onto suitable subspaces. The algorithm is tested on a simulated data set.
Keywords :
belief networks; computational complexity; interactive programming; learning systems; multiprogramming; Bayesian online multitask learning; computational complexity; regularization networks; single-task kernel; Bayesian methods; Computational complexity; Filtering; Kalman filters; Kernel; Machine learning; Machine learning algorithms; Random variables; Recursive estimation; Testing; Bayesian estimation; Kalman filtering; kernel methods; machine learning; multi-task learning; regularization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
American Control Conference, 2008
Conference_Location :
Seattle, WA
ISSN :
0743-1619
Print_ISBN :
978-1-4244-2078-0
Electronic_ISBN :
0743-1619
Type :
conf
DOI :
10.1109/ACC.2008.4587207
Filename :
4587207
Link To Document :
بازگشت