Decentralized Estimation using distortion sensitive learning vector quantization

Nonnegative Matrix Factorization using Class Label Information

“…To combine the the class label information with NMF, we introduce the following matrix so that = (5) So that if and belong to the same class, there representations and will be the same, as shown in Figure. 1. Based on the new representation of and the squared Euclidean distance loss function function, our NMF-CLI algorithm with the class label information is reduced the following optimization problem:…”

“…Matrix factorization technique is having attention as a good method to do data representation [2]. Various methods of representation have been introduced by applying different formulas, including Singular Value Decomposition [3], Principal Component Analysis [4], and Vector Quantization [5]. Matrix factorization means seeking two matrix factors, such that their product can approximate the matrix [6].…”

Adaptive conformal semi-supervised vector quantization for dissimilarity data

“…In contrast with the black box property of SVM and its semi-supervised variants, prototype-based methods enjoy a wide popularity in various application domains (Grbovic and Vucetic, 2013;Ortiz-Bayliss et al, 2013) due to their intuitive and simple behavior: they represent their decision in terms of typical representatives (referred to as prototypes) in the input space and classification is based on the distances of data to prototypes. Prototypes can be directly inspected by domain experts in the field in the same way as data points.…”