Semi-supervised learning in exemplar based neural networks
Keywords
Neural networks (Computer science); Supervised learning (Machine learning)
Abstract
A new and novel method for training exemplar based neural networks called Semisupervised learning is introduced. Semi-supervised learning (SSL) refers to the semisupervised manner according to which exemplars are formed during training to identify clusters. Instead of forcing a neural network to achieve a zero post training error a new learning scheme is introduced that would allow exemplars to occasionally misclassify training patterns by permitting training patterns, under certain circumstances, to modify exemplars associated with a different class. This allows the network to achieve a smooth learning curve without being disrupted by the noise in the dataset. The new technique was applied to 3 exemplar-based classifiers and results were found to be very promising. The new semi-supervised networks performed better classification and achieved better compression when compared to their supervised counterparts.
Notes
This item is only available in print in the UCF Libraries. If this is your thesis or dissertation, you can help us make it available online for use by researchers around the world by STARS for more information.
Graduation Date
2003
Advisor
Georgiopoulos, Michael
Degree
Master of Science (M.S.)
College
College of Engineering
Department
Electrical Engineering and Computer Science
Format
Pages
200 p.
Language
English
Length of Campus-only Access
None
Access Status
Masters Thesis (Open Access)
Identifier
DP0029106
Subjects
Dissertations, Academic -- Engineering; Engineering -- Dissertations, Academic
STARS Citation
Bharadwaj, Madan, "Semi-supervised learning in exemplar based neural networks" (2003). Retrospective Theses and Dissertations. 761.
https://stars.library.ucf.edu/rtd/761