International Workshop on Pattern Recognition in NeuroImaging

Many modeling techniques have been developed to try to recover the maximum available information from fMRI data. One approach that has become increasingly popular is voxel-wise modeling and decoding (VWMD). Under this approach each voxel is modeled separately by estimating the transformation between a set of hypothetical stimulus or task features and measured BOLD responses. Different hypothetical feature spaces can be tested by comparing model predictions. Feature tuning in individual voxels can be assessed, and tuning patterns across voxels can be determined by clustering or dimensionality reduction. Decoding can be performed by using the estimated encoding models to compute the multi-voxel likelihood, and then combining this with an appropriate prior to obtain the posterior distribution over stimuli. Thus, VWMD shares many properties of other pattern analysis techniques used currently to analyze fMRI data. However, VWMD does have a few specific advantages over competing approaches. Most importantly, it permits analysis of complex, continuous experimental designs that would be extremely difficult to assess using other methods. In this talk I will summarize the VWMD approach used in my laboratory to understand how the brain represents information in continuous natural movies. I will address model selection, fitting and evaluation; interpretation of the fit models; and movie reconstruction (decoding).

Jack Gallant is Professor of Psychology at the University of California at Berkeley, and is affiliated with the graduate programs in Bioengineering, Biophysics, Neuroscience and Vision Science. He received his Ph.D. from Yale University and did post-doctoral work at the California Institute of Technology and Washington University Medical School. His research program focuses on constructing quantitative computational models that accurately describe how the brain encodes information during natural tasks, and to use these models to decode information in the brain in order to reconstruct mental experiences. This computational framework can be used to understand and decode brain activity measured by different methods (e.g., functional MRI, NIRS, EEG or ECOG), and in different modalities (i.e., vision, audition, imagery and so on).

Human neuroimaging promises to provide new answers to fundamental questions about mind and brain. The field has taken some promising steps towards this goal, but there are fundamental limitations in the way in which we have used data to address questions about the mind. A new way of using fMRI data, a "Neuroimaging 2.0," can move us closer to the solution, and machine learning is an integral part of that endeavor. The approach I will describe is grounded in a) explaining phenomena of direct psychological, medical, or philosophical interest; b) using data within individuals, across individuals, and across many studies to define explicit biomarkers for such phenomena; and c) cumulative, quantitative validation of such biomarkers for particular classes of mental events. I will use examples from affective neuroscience to illustrate how machine learning techniques can be used in the service of answering psychological questions about the nature of pain and emotion and how they are influenced by cognitive processes.

Kernel methods in machine learning have expanded from tricks to construct nonlinear algorithms to general tools to assay higher order statistics and properties of distributions. They find applications also in causal inference, an intriguing field that examines causal structures by testing their probabilistic footprints. However, the links between causal inference and machine learning go beyond this, and the talk will outline a few thoughts how some challenging problems of modern machine learning can benefit from the causal methodology.

Bernhard Schölkopf was born in Stuttgart on 20 February, 1968. He received an M.Sc. in mathematics and the Lionel Cooper Memorial Prize from the University of London in 1992, followed in 1994 by the Diplom in physics from the Eberhard-Karls-Universität, Tübingen. Three years later, he obtained a doctorate in computer science from the Technical University Berlin. His thesis on Support Vector Learning won the annual dissertation prize of the German Association for Computer Science (GI). In 1998, he won the prize for the best scientific project at the German National Research Center for Computer Science (GMD). He has researched at AT&T Bell Labs, at GMD FIRST, Berlin, at the Australian National University, Canberra, and at Microsoft Research Cambridge (UK). He has taught at Humboldt University, Technical University Berlin, and Eberhard-Karls-University Tübingen. In July 2001, he was appointed scientific member of the Max Planck Society and director at the MPI for Biological Cybernetics; in October 2002, he was appointed Honorarprofessor for Machine Learning at the Technical University Berlin. In 2006, he received the J. K. Aggarwal Prize of the International Association for Pattern Recognition, in 2011, he got the Max Planck Research Award. The ISI lists him as a highly cited researcher. He served on the editorial boards of JMLR, IEEE PAMI, and IJCV.