A NOVEL NEUTROSOPHIC LOGIC SVM (N-SVM) AND ITS APPLICATION TO IMAGE CATEGORIZATION

摘要: Neutrosophic logic is a relatively new that generalization of fuzzy logic. In this paper, for the first time, neutrosophic applied to field classifiers where support vector machine (SVM) adopted as example validate its feasibility and effectiveness. The proposed set integrated into reformulated SVM, performance obtained classifier N-SVM evaluated under region-based image categorization system. Images are segmented by hierarchical two-stage self-organizing map (HSOM) using color texture features. A novel approach select training samples HSOM based on homogeneity properties. diverse density (DD-SVM) framework then viewing an bag instances corresponding regions from segmentation. Each mapped point in space, transformed classification problem. Then, used space. treats differently according weighting function, it helps reduce effects outliers. Experimental results have demonstrated validity effectiveness method which may find wide applications related areas.