Inter-subject alignment is an important aspect of multisubject fMRI research. Recently a method known as Hyperalignment has shown considerable success in attaining such alignment. In order to improve computational efficiency, we investigate a joint SVD-Hyperalignment algorithm. We show that this algorithm is more scalable than the standard Hyperalignment algorithm by providing analytic and empirical results using a multi-subject fMRI dataset. The experimental results show improved computation speed while maintaining between subject prediction accuracy on an image viewing experiment. In addition, our results provide benchmark relationships between voxel selection, accuracy and computation complexity for Hyperalignment, taking a joint SVD of the data, and joint SVD-Hyperalignment.

Several research groups have shown how to map fMRI responses to the meanings of presented stimuli. This paper presents new methods for doing so when only a natural language annotation is available as the description of the stimulus. We study fMRI data gathered from subjects watching an episode of BBCs Sherlock [1], and learn bidirectional mappings between fMRI responses and natural language representations. By leveraging data from multiple subjects watching the same movie, we were able to perform scene classification with 72% accuracy (random guessing would give 4%) and scene ranking with average rank in the top 4% (random guessing would give 50%). The key ingredients underlying this high level of performance are (a) the use of the Shared Response Model (SRM) and its variant SRM-ICA [2, 3] to aggregate fMRI data from multiple subjects, both of which are shown to be superior to standard PCA in producing low-dimensional representations for the tasks in this paper; (b) a sentence embedding technique adapted from the natural language processing (NLP) literature [4] that produces semantic vector representation of the annotations; (c) using previous timestep information in the featurization of the predictor data. These optimizations in how we featurize the fMRI data and text annotations provide a substantial improvement in classification performance, relative to standard approaches.