Structured outputs such as multidimensional vectors or graphs are frequently encountered in real world pattern recognition applications such as computer vision, natural language pr...
Many tasks in speech processing involve classification of long term characteristics of a speech segment such as language, speaker, dialect, or topic. A natural technique for dete...
We present a discriminative training algorithm, that uses support vector machines (SVMs), to improve the classification of discrete and continuous output probability hidden Markov ...
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...
A Half-Against-Half (HAH) multi-class SVM is proposed in this paper. Unlike the commonly used One-Against-All (OVA) and One-Against-One (OVO) implementation methods, HAH is built ...