Multi-class support vector machines for protein secondary structure prediction.

摘要: The solution of binary classification problems using the Support Vector Machine (SVM) method has been well developed. Though multi-class is typically solved by combining several classifiers, recently, methods that consider all classes at once have proposed. However, these require resolving a much larger optimization problem and are applicable to small datasets. Three based on classifications: one-against-all (OAA), one-against-one (OAO), directed acyclic graph (DAG), two approaches for solving one single problem, implemented predict protein secondary structure. Our experiments indicate SVM more suitable structure (PSS) prediction than other methods, including SVMs, because their capacity solve an in step. Furthermore, this paper, we argue it feasible extend accuracy adding second-stage capture contextual information among structural elements thereby further improving accuracies. We demonstrate two-stage SVMs perform better single-stage techniques PSS datasets report maximum 79.5%.