We discuss the problem of overfitting of probabilistic neural networks in the framework of statistical pattern recognition. The probabilistic approach to neural networks provides a...
Abstract-- Many statistical measures and algorithmic techniques have been proposed for studying residue coupling in protein families. Generally speaking, two residue positions are ...
John Thomas, Naren Ramakrishnan, Chris Bailey-Kell...
Background: Identifying quantitative trait loci (QTL) for both additive and epistatic effects raises the statistical issue of selecting variables from a large number of candidates...
The conditional distribution of a discrete variable y, given another discrete variable x, is often specified by assigning one multinomial distribution to each state of x. The cost...
Background: Multicategory Support Vector Machines (MC-SVM) are powerful classification systems with excellent performance in a variety of data classification problems. Since the p...