Date of Submission


Date of Award


Institute Name (Publisher)

Indian Statistical Institute

Document Type

Doctoral Thesis

Degree Name

Doctor of Philosophy

Subject Name

Computer Science


Machine Intelligence Unit (MIU-Kolkata)


Mitra, Sushmita (MIU-Kolkata; ISI)

Abstract (Summary of the Work)

Pattern recognition is what humans do most of the time, without any conscious effort, and fortunately excel in. Information is received through various sensory organs, processed simultaneously in the brain, and its source is instantaneously identified without any perceptible effort. The interesting issue is that recognition occurs even under non-ideal conditions, i.e., when information is vague, imprecise or incomplete. In reality, most human activities depend on the success in performing various pattern recognition tasks. Let us consider an example. Before boarding a train or bus, we first select the appropriate one by identifying either the route number or its destination on the basis of the visual signals received by the brain; this information is then speedily processed, followed by neurobiological implementation of template-matching.The objective of this thesis is to present development and design of some algorithms, along with their case studies, involving both theoretical and experimental studies in unsupervised feature selection. Extension to large data is also investigated, with a view to reducing the curse of dimensionality. Novel similarity measures, from statistical, classical, and soft computing domains, are introduced to identify reduced subsets of informative features. The similarity is mainly based on various internal characteristics of the data.


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.


Included in

Mathematics Commons