Instance-based Cost Sensitive Boosting

Many classification algorithms aim to minimize just their training error count; however, it is often desirable to minimize a more general cost metric, where distinct instances have different costs....