《Deep learning for neuroimaging a validation study》.pdfVIP

Deep learning for neuroimaging: a validation study Sergey M. Plis Devon Hjelm The Mind Research Network University of New Mexico Albuquerque, NM 87106 Albuquerque, NM 87131 s.m.plis@ dhjelm@ 3 1 0 Ruslan Salakhutdinov Vince D. Calhoun 2 University of Toronto The Mind Research Network Toronto, Ontario M5S 2E4 Albuquerque, NM 87106 c e rsalakhu@ vcalhoun@ D 0 2 Abstract ] Deep learning methods have recently enjoyed a number of successes in the tasks E of classification and representation learning. These tasks are very important for N brain imaging and neuroscience discovery, making the methods attractive candi- s. dates for porting to a neuroimager’s toolbox. Successes are, in part, explained c by a great flexibility of deep learning models. This flexibility makes the process [ of porting to new areas a difficult parameter optimization problem. In this work 1 we demonstrate our results (and feasible parameter ranges) in application of deep v learning methods to structural and functional brain imaging data. We also describe 7 a novel constraint-based approach to visualizing high dimensi