Learning Neural Representations of Human Cognition across Many fMRI Studies

Abstract : Cognitive neuroscience is enjoying rapid increase in extensive public brain-imaging datasets. It opens the door to large-scale statistical models. Finding a unified perspective for all available data calls for scalable and automated solutions to an old challenge: how to aggregate heterogeneous information on brain function into a universal cognitive system that relates mental operations/cognitive processes/psychological tasks to brain networks? We cast this challenge in a machine-learning approach to predict conditions from statistical brain maps across different studies. For this, we leverage multi-task learning and multi-scale dimension reduction to learn low-dimensional representations of brain images that carry cognitive information and can be robustly associated with psychological stimuli. Our multi-dataset classification model achieves the best prediction performance on several large reference datasets, compared to models without cognitive-aware low-dimension representations; it brings a substantial performance boost to the analysis of small datasets, and can be introspected to identify universal template cognitive concepts.
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01626823
Contributor : Arthur Mensch <>
Submitted on : Saturday, November 11, 2017 - 4:27:02 AM
Last modification on : Friday, March 8, 2019 - 1:20:18 AM
Document(s) archivé(s) le : Monday, February 12, 2018 - 12:16:38 PM

Files

nips.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01626823, version 3
  • ARXIV : 1710.11438

Citation

Arthur Mensch, Julien Mairal, Danilo Bzdok, Bertrand Thirion, Gaël Varoquaux. Learning Neural Representations of Human Cognition across Many fMRI Studies. Neural Information Processing Systems, Dec 2017, Long Beach, United States. ⟨hal-01626823v3⟩

Share

Metrics

Record views

1328

Files downloads

299