Performance Evaluation of Kernel-Based Supervised Noise Clustering Approach

Examination of Kernel Based Noise Classifier with Cross Resolution Dataset and with Untrained Class

“…This paper is the extension of the previous work done related kernel based noise classifier, KNC, using nine-kernel function in supervised model [16,17]. The objective of present paper is to assess the associativity of untrained class upon kernel based noise classifier.…”

Assimilation of Standard Regularizer Contextual Model and Composite Kernel with Fuzzy-based Noise Classifier

“…This approach has been used to estimate the performance of fuzzy based classifier and also to spot classifier conduct with the mixed pixels [16,17,20]. Class wise sample data or generation of this image has been prepared from mean vector of the classes via distance measure.…”

“…Class wise sample data or generation of this image has been prepared from mean vector of the classes via distance measure. The image has been partitioned into three variations -Pure Pixel composition, Mixed Pixel Composition (50:50) between two classes, and Mixed Pixel Composition (30:30:40) among three classes [20]. Undergoing supervised classification, the pixel values will be examined with the aimed membership value of 0.50, 0.40 and 0.30 as anticipated pixel with 50%, 40% and 30% fitting to the class respectively.…”

“…The kernels based on distance function are local kernels [20,21]. Gaussian kernel is a form of local kernel, where is the feature vector in the image and y is the mean vector of the class.…”

“…Sigmoid kernel is a hyperbolic tangent function and belongs to the class of global kernel [21]. The scaling parameter is α for the kernel functions that defines width of the kernel [20].…”