Robust multiple-instance learning ensembles using random subspace instance selection
Many real-world pattern recognition problems can be modeled using multiple-instance learning (MIL), where instances are grouped into bags, and each bag is assigned a label. State

