Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Multi-disciplinary Studies | en_US |
| dc.creator | Chiu, Man-yau | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/2173 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | - |
| dc.rights | All rights reserved | en_US |
| dc.title | Study of neuro-fuzzy classifiers with an application to language separation of textual images | en_US |
| dcterms.abstract | In this project, we developed a neuro-fuzzy classifier, which combines the merits of both fuzzy logic and neural networks, in order to achieve 'human-like' performance in pattern recognition. This neuro-fuzzy classifier can be trained using a set of training patterns by a hybrid learning scheme. Firstly, the K-means clustering is performed on the training patterns to initialize membership function parameters and to generate a set of fuzzy rules. Then, the parameters of the neuro-fuzzy classifier, namely membership function parameters, aggregator's parameters, and defuzzification parameters, are fine-tuned by using the error-backpropagation training algorithm to meet the pre-defined convergence criteria. By combining both unsupervised (K-means clustering) and supervised (error-backpropagation training algorithm) learning schemes, we can achieve a learning process which is faster than the one by using the error-backpropagation training algorithm alone. Furthermore, incorporating the fuzzy reasoning into a neural network model not only eliminates a disadvantage of a conventional multi-layer feedforward neural network in understanding and predicting its performance but also achieves an efficient and high performance defuzzification. The designed neuro-fuzzy classifier has been used to perform the textual image recognition task in which Chinese and English textual images are classified. In order to evaluate the effect of different network settings on the classifier's performance, we have tested different number of fuzzy partition for each input variable (3 or 5 linguistic labels) and different types of aggregation operators, namely local parameterized mean operator, global parameterized mean operator, and product operator, in constructing the neuro-fuzzy classifier. Experimental results show that a neuro-fuzzy classifier with each of the three aggregation operators can successfully classify all the training and the test textual images into correct language classes. | en_US |
| dcterms.extent | v, 139 leaves : ill. ; 30 cm | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 1998 | en_US |
| dcterms.educationalLevel | All Master | en_US |
| dcterms.educationalLevel | M.Sc. | en_US |
| dcterms.LCSH | Fuzzy systems | en_US |
| dcterms.LCSH | Neural networks (Computer science) | en_US |
| dcterms.LCSH | Image processing -- Digital techniques | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.accessRights | restricted access | en_US |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| b14211282.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 3.89 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/2173