"The problem of clustering is one of the most widely studied area in data mining and machine learning. Adaptive resonance theory (ART), an unsupervised learning clustering algorithm, is a clustering method that can learn arbitrary input patterns in a stable, fast and self-organizing way. This dissertation focuses on unsupervised learning methods, mostly based on variations of ART.
Hierarchical ART clustering is studied by generating a tree of ART units with GPU based parallelization to provide fast and finesse clustering. Experiment results show that the our method achieves significant training speed increase in generating deep ART trees compared with that from non-parallelized version.
In order to handle high dimensional, noisy data more accurately, a hierarchical biclustering ARTMAP (H-BARTMAP) is developed. The nature of biclustering, which considers the correlation of each members in clusters, combined with the concept of hierarchical clustering, provides highly accurate experimental results, especially in bioinformatics data sets.
The third paper focuses on applying the biclustering concept to a supervised learning method, named supervised BARTMAP (S-BARTMAP). Experimental results on high dimensional data sets show that S-BARTMAP is capable of making better predictions compared with those from other math based and machine learning methods
The final paper focuses on solving the semi-supervised support vector machine (S3VM) optimization problem with the aid of value gradient learning (VGL). By applying a reinforcement learning method to a semi-supervised problem results in a solid classification performance in terms of cluster validation, better than algorithms from previous studies"--Abstract, page iv.
Wunsch, Donald C.
Sarangapani, Jagannathan, 1965-
Electrical and Computer Engineering
Ph. D. in Computer Engineering
National Science Foundation (U.S.)
United States. Army Research Office
Missouri University of Science and Technology. Intelligent Systems Center
Mary K. Finley Missouri Endowment
Missouri University of Science and Technology
Journal article titles appearing in thesis/dissertation
- A GPU based parallel hierarchical fuzzy ART clustering
- Hierarchical BARTMAP: A novel innovation in biclustering algorithms
- Value Gradient Learning based Semi-Supervised Support Vector machines
- Biclustering based prediction with supervised biclustering ARTMAP
xiii, 111 pages
© 2016 Sejun Kim, All rights reserved.
Dissertation - Open Access
Library of Congress Subject Headings
Pattern recognition systems
Electronic OCLC #
Kim, Sejun, "Novel approaches to clustering, biclustering algorithms based on adaptive resonance theory and intelligent control" (2016). Doctoral Dissertations. 2478.