Dynamic guard zone for self-supervised learning

Pal, S. K. ; Pathak, A. ; Basu, C. (1988) Dynamic guard zone for self-supervised learning Pattern Recognition Letters, 7 (3). pp. 135-144. ISSN 0167-8655

Full text not available from this repository.

Official URL: http://linkinghub.elsevier.com/retrieve/pii/016786...

Related URL: http://dx.doi.org/10.1016/0167-8655(88)90056-6


The dimension of the guard zone along with its bounds for the Generalised Learning Algorithm (Pathak and Pal, 1986) is determined for optimum learning. The dimension is found to be dynamic depending on the input sequence and the current estimates of classification parameters. Incorporation of this higher-order knowledge in a supervisory program improves the system performance. The performance is again found to be affected if the guard zone is shrunk/expanded for 'very weak'/'not too weak' estimates when speech data is considered to be input.

Item Type:Article
Source:Copyright of this article belongs to International Association for Pattern Recognition.
Keywords:Dynamic Guard Zone; Learning; Speech Recognition
ID Code:26131
Deposited On:06 Dec 2010 13:02
Last Modified:13 Jun 2011 06:11

Repository Staff Only: item control page