Musk2 Data Set

Description

The problem consists of determining whether a drug molecule will bind strongly to a target protein. Each molecule may adopt a wide range of shapes or conformations. A positive molecule has at least one shape that can bind well (although it is not known which one) and a negative molecule means none of its shapes can make the molecule bind well. This problem could be represented in a very natural way in MIL settings: each molecule would be a bag and the conformations it can adopt would be the instances in that bag.

Dataset

The original data set is partitioned using 10-fold cross-validation procedure five times. Thus, five different partitions of 10-fold cross validation are available