Boosting interval based literals1

Abstract

A supervised classification method for time series, even multivariate, is presented. It is based on boosting very simple classifiers: clauses with one literal in the body. The background predicates are based on temporal intervals. Two types of predic