The increasing of public neuroimaging datasets opens a door to analyzing homogeneous human brain conditions across datasets by transfer learning (TL). However, neuroimaging data are high-dimensional, noisy, and with small sample sizes. It is challenging to learn a robust model for data across different cognitive experiments and subjects. A recent TL approach minimizes domain dependence to learn common cross-domain features, via the Hilbert-Schmidt Independence Criterion (HSIC). Inspired by this approach and the multi-source TL theory, we propose a Side Information Dependence Regularization (SIDeR) learning framework for TL in brain condition decoding. Specifically, SIDeR simultaneously minimizes the empirical risk and the statistical dependence on the domain side information, to reduce the theoretical generalization error bound. We construct 17 brain decoding TL tasks using public neuroimaging data for evaluation. Comprehensive experiments validate the superiority of SIDeR over ten competing methods, particularly an average improvement of 15.6% on the TL tasks with multi-source experiments.

Representational Similarity Learning (RSL) aims to discover features that are important in representing (human-judged) similarities among objects. RSL can be posed as a sparsity-regularized multi-task regression problem. Standard methods, like group lasso, may not select important features if they are strongly correlated with others. To address this shortcoming we present a new regularizer for multitask regression called Group Ordered Weighted \ell_1 (GrOWL). Another key contribution of our paper is a novel application to fMRI brain imaging. Representational Similarity Analysis (RSA) is a tool for testing whether localized brain regions encode perceptual similarities. Using GrOWL, we propose a new approach called Network RSA that can discover arbitrarily structured brain networks (possibly widely distributed and non-local) that encode similarity information. We show, in theory and fMRI experiments, how GrOWL deals with strongly correlated covariates.

Multitask learning can be effective when features useful in one task are also useful for other tasks, and the group lasso is a standard method for selecting a common subset of features. In this paper, we are interested in a less restrictive form of multitask learning, wherein (1) the available features can be organized into subsets according to a notion of similarity and (2) features useful in one task are similar, but not necessarily identical, to the features best suited for other tasks. The main contribution of this paper is a new procedure called {\em Sparse Overlapping Sets (SOS) lasso}, a convex optimization that automatically selects similar features for related learning tasks. Error bounds are derived for SOSlasso and its consistency is established for squared error loss. In particular, SOSlasso is motivated by multi-subject fMRI studies in which functional activity is classified using brain voxels as features. Experiments with real and synthetic data demonstrate the advantages of SOSlasso compared to the lasso and group lasso.