Hierarchical Region-Network Sparsity for High-Dimensional Inference in Brain Imaging

Abstract : Structured sparsity penalization has recently improved statistical models applied to high-dimensional data in various domains. As an extension to medical imaging, the present work incorporates priors on network hierarchies of brain regions into logistic-regression to distinguish neural activity effects. These priors bridge two separately studied levels of brain architecture: functional segregation into regions and functional integration by networks. Hierarchical region-network priors are shown to better classify and recover 18 psychological tasks than other sparse esti-mators. Varying the relative importance of region and network structure within the hierarchical tree penalty captured complementary aspects of the neural activity patterns. Local and global priors of neurobiological knowledge are thus demonstrated to offer advantages in generalization performance, sample complexity, and domain interpretability.
Liste complète des métadonnées

Cited literature [22 references]  Display  Hide  Download

Contributor : Danilo Bzdok <>
Submitted on : Wednesday, March 1, 2017 - 7:38:59 PM
Last modification on : Friday, March 8, 2019 - 1:20:16 AM
Document(s) archivé(s) le : Tuesday, May 30, 2017 - 6:30:38 PM


Files produced by the author(s)


  • HAL Id : hal-01480885, version 1


Danilo Bzdok, Michael Eickenberg, Gaël Varoquaux, Bertrand Thirion. Hierarchical Region-Network Sparsity for High-Dimensional Inference in Brain Imaging. International conference on Information Processing in Medical Imaging (IPMI) 2017, Jun 2017, Boone, North Carolina, United States. ⟨hal-01480885⟩



Record views


Files downloads