학술논문
Type-2 fuzzy neural network for lip activity detection
Document Type
Conference
Author
Source
2017 International Conference on Fuzzy Theory and Its Applications (iFUZZY) Fuzzy Theory and Its Applications (iFUZZY), 2017 International Conference on. :1-4 Nov, 2017
Subject
Language
ISSN
2377-5831
Abstract
A type-2 fuzzy neural network (T2FNN) is proposed for lip activity detection (LAD). In image and speech signal processing, LAD is an important classification problem. The gradient descent method is applied to reduce the cost function of T2FNN. In handing noisy data with uncertainties, type-2 fuzzy-systems usually outperform their type-1 counterparts. By structure and parameter learning, the classification of T2FNN has high discriminability. Compared with other existing fuzzy neural networks, the advantage of the T2FNN is its consideration of uncertainty. The effectiveness of the T2FNN is tested by lip activity detection. Experimental result shows that T2FNN works better than the other type-1 fuzzy neural networks.