Pattern Synthesis with Active Learning Using Classifier Combination.

Date of Submission

December 2005

Date of Award

Winter 12-12-2006

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science


Machine Intelligence Unit (MIU-Kolkata)


Ghosh, Ashish (MIU-Kolkata; ISI)

Abstract (Summary of the Work)

In this dissertation, the problem of Pattern Synthesis in conjunction with Ac- tive Learning has been considered. Synthesising new patterns though initially appears to be hardly possible, the research community has been dealing with the inverse problem i.e., the data condensation problem fairly well. Based on this idea a novel pattern synthesis technique mostly mimicing the DBSCAN clustering algorithm has been proposed. The patterns crrated in this way have been used to the Active Learning process. Moreover, it is well known that in many situations a Classifier Combination, consistently outperforms a single best classifier, and when there are a number of classifiers of different nature, one can use the Query By Committee(QBC) approach for actively selecting the new patterns for further training. In this dissertation Active Learning using Classifier Combination coupled with Pattern Synthesis has been studied. Experimental evaluation also reveals that this is slightly better than the naive Passive Learning approach using Classifier Combination. In this dissertation, the problem of Pattern Synthesis in conjunction


ProQuest Collection ID:

Control Number


Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.


This document is currently not available here.