Feature Induction Using Boosting and Logistic Regression on fMRI Images

Early efforts in fMRI classification were limited in that individual voxels were used as features (e.g. [1]), yet voxels divide images into regions that do not directly correspond to underlying neur