Academic Journals Database
Disseminating quality controlled scientific knowledge

An incremental learning algorithm considering texts' reliability

Author(s): Xinghua Fan | Shaozhu Wang

Journal: International Journal of Advanced Computer Sciences and Applications
ISSN 2156-5570

Volume: 3;
Issue: 2;
Start page: 23;
Date: 2012;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: text classification | incremental learning | reliability | text distribution | evaluation.

The sequence of texts selected obviously influences the accuracy of classification. Some sequences may make the performance of classification poor. For overcoming this problem, an incremental learning algorithm considering texts’ reliability, which finds reliable texts and selects them preferentially, is proposed in this paper. To find reliable texts, it uses two evaluation methods of FEM and SEM, which are proposed according to the text distribution of unlabeled texts. The results of the last experiments not only verify the effectiveness of the two evaluation methods but also indicate that the proposed incremental learning algorithm has advantages of fast training speed, high accuracy of classification, and steady performance.

Tango Jona
Tangokurs Rapperswil-Jona

     Affiliate Program