Doctoral Dissertations

Keywords and Phrases

Adaptive Resonance Theory; Clustering; Cluster Validity Index; Data Visualization; Neural Networks; Self-Organizing Maps

Abstract

"Cluster analysis can be broadly divided into multivariate data visualization, clustering algorithms, and cluster validation. This dissertation contributes neural network-based techniques to perform all three unsupervised learning tasks. Particularly, the first paper provides a comprehensive review on adaptive resonance theory (ART) models for engineering applications and provides context for the four subsequent papers. These papers are devoted to enhancements of ART-based clustering algorithms from (a) a practical perspective by exploiting the visual assessment of cluster tendency (VAT) sorting algorithm as a preprocessor for ART offline training, thus mitigating ordering effects; and (b) an engineering perspective by designing a family of multi-criteria ART models: dual vigilance fuzzy ART and distributed dual vigilance fuzzy ART (both of which are capable of detecting complex cluster structures), merge ART (aggregates partitions and lessens ordering effects in online learning), and cluster validity index vigilance in fuzzy ART (features a robust vigilance parameter selection and alleviates ordering effects in offline learning). The sixth paper consists of enhancements to data visualization using self-organizing maps (SOMs) by depicting in the reduced dimension and topology-preserving SOM grid information-theoretic similarity measures between neighboring neurons. This visualization's parameters are estimated using samples selected via a single-linkage procedure, thereby generating heatmaps that portray more homogeneous within-cluster similarities and crisper between-cluster boundaries. The seventh paper presents incremental cluster validity indices (iCVIs) realized by (a) incorporating existing formulations of online computations for clusters' descriptors, or (b) modifying an existing ART-based model and incrementally updating local density counts between prototypes. Moreover, this last paper provides the first comprehensive comparison of iCVIs in the computational intelligence literature"--Abstract, page iv.

Advisor(s)

Wunsch, Donald C.

Committee Member(s)

Moss, Randy Hays, 1953-
Stanley, R. Joe
Zhang, Jiangfan
Dagli, Cihan H., 1949-

Department(s)

Electrical and Computer Engineering

Degree Name

Ph. D. in Computer Engineering

Comments

This research was sponsored by the Missouri University of Science and Technology Mary K. Finley Endowment and Intelligent Systems Center; the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brazil (CAPES) - Finance code BEX 13494/13-9; the Army Research Laboratory (ARL) and the Lifelong Learning Machines program from DARPA/MTO, and it was accomplished under Cooperative Agreement Number W911NF- 18-2-0260.

Research Center/Lab(s)

Intelligent Systems Center

Publisher

Missouri University of Science and Technology

Publication Date

Fall 2019

Journal article titles appearing in thesis/dissertation

  • A survey of adaptive resonance theory neural network models for engineering applications
  • Validity index-based vigilance test in adaptive resonance theory neural networks
  • A study on exploiting VAT to mitigate ordering effects in fuzzy ART
  • Dual vigilance fuzzy adaptive resonance theory
  • Distributed dual vigilance fuzzy adaptive resonance theory learns online, retrieves arbitrarily-shaped clusters, and mitigates order dependence
  • An information-theoretic-cluster visualization for self-organizing maps
  • Incremental cluster validity indices for hard partitions: Extensions and comparative study

Pagination

xxv, 407 pages

Note about bibliography

Includes bibliographic references.

Rights

© 2019 Leonardo Enzo Brito da Silva, All rights reserved.

Document Type

Dissertation - Open Access

File Type

text

Language

English

Thesis Number

T 11616

Electronic OCLC #

1139525590

Download

Share

 
COinS