Feature Space Interpretation of SVMs with Indefinite Kernels
IEEE Transactions on Pattern Analysis and Machine Intelligence
Using One-Class and Two-Class SVMs for Multiclass Image Annotation
IEEE Transactions on Knowledge and Data Engineering
Feature selection in MLPs and SVMs based on maximum output information
IEEE Transactions on Neural Networks
An improved conjugate gradient scheme to the solution of least squares SVM
IEEE Transactions on Neural Networks
Hi-index | 0.00 |
The probability model is introduced into classification learning in this paper. Kernel covering algorithm (KCA) and maximum likelihood principle of the statistic model combine to form a novel algorithm-the probability model of covering algorithm (PMCA) which not only introduces optimization processing for every covering domain, but offers a new way to solve the parameter problem of kernel function. Covering algorithm (CA) is firstly used to get covering domains and approximate interfaces, and then maximum likelihood principle of finite mixture model is used to fit each distributing. Experiments indicate that optimization is surely achieved, classification rate is improved and the neural cells are cut down greatly through with proper threshold value.