An analog system-on-chip for kernel-based pattern classification and sequence estimation is presented. State transition probabilities conditioned on input data are generated by an...
A binary hierarchical classifier is proposed to solve the multi-class classification problem. We also require rejection of non-target inputs, which thus producing a very difficult...
—Emerging applications of computer vision and pattern recognition in mobile devices and networked computing require the development of resourcelimited algorithms. Linear classifi...
Abstract. We report about some experiments on the fingerprint database NIST-4 using different combinations of Support Vector Machine (SVM) classifiers. Images have been preprocesse...
In this paper, we propose a method for support vector machine classification using indefinite kernels. Instead of directly minimizing or stabilizing a nonconvex loss function, our...