The thesis put forth in this dissertation is that machine learning classifiers can be used as instruments for decoding variables of interest from functional magnetic resonance imaging (ƒMRI) data. There are two main goals in decoding:

• Showing that the variable of interest can be predicted from the data in a statistically reliable manner (i.e. there's enough information present).
• Shedding light on how the data encode the information needed to predict, taking into account what the classifier used can learn and any criteria by which the data are filtered (e.g. how voxels and time points used are chosen).

Chapter 2 considers the issues that arise when using traditional linear classifiers and several different voxel selection techniques to strive towards these goals. It examines questions such as whether there is redundant, as well as different kinds of, information in ƒMRI data and how those facts complicate the task of determining whether voxel subsets encode the desired information or whether certain choices of selection method or classifier are universally preferrable. All the results presented were obtained through a comparative study of five ƒMRI datasets.

Chapter 3 and Chapter 4 introduce the Support Vector Decomposition Machine, a new algorithm that attempts to sidestep the use of voxel selection by learning a low dimensional representation of the data that is also suitable for decoding variables of interest and has the potential to allow incorporation of domain information directly, rather than through the proxy of voxel selection criteria.

212 pages

Return to: SCS Technical Report Collection
School of Computer Science

This page maintained by [email protected]