Input selection for linear and nonlinear modelling is an important problem, related to the trade-off between model complexity and in sample model accuracy. For linear modelling, well-known complexity criteria like the Akaike and Bayesian Information Criteria have been developed. In this paper, we explain the Bayesian evidence framework for Least Squares Support Vector Machines (LS-SVMs) and explain its use for input selection. bayesian learning; input selection; kernel based learning; least squares support vector machines; nonlinear regression
展开▼
