• Home
• Research
• Personal

research

what's in the box?

• Can we train classifiers to decode a variable of interest, such as what a subject is seeing, thinking or what decision she is about to make, from images acquired at the time the thoughts took place?
• Can we use successful classifiers to tell us what, in the spatiotemporal pattern of activation, is connected to that variable and how it is connected?
• How can we bring existing knowledge in cognitive neuroscience to bear as constraints when learning classifiers, so that they learn models of the data that are useful to someone generating or testing hypotheses about the cognitive process being studied?

publications

• "The Support Vector Decomposition Machine"
  
  Pereira F., Gordon G.
  to appear in International Conference on Machine Learning, 2006
  [abstract] [paper]
  

  In machine learning problems with tens of thousands of features and only dozens or hundreds of independent training examples, dimensionality reduction is essential for good learning performance. In previous work, many researchers have treated the learning problem in two separate phases: first use an algorithm such as singular value decomposition to reduce the dimensionality of the data set, and then use a classification algorithm such as naive Bayes or support vector machines to learn a classifier. We demonstrate that it is possible to combine the two goals of dimensionality reduction and classification into a single learning objective, and present a novel and efficient algorithm which optimizes this objective directly. We present experimental results in fMRI analysis which show that we can achieve better learning performance and lower-dimensional representations than two-phase approaches can.
• "Exploring predictive and reproducible modeling with the single-subject FIAC data set"
  
  Chen X., Pereira F., Lee W., Strother S., Mitchell T.
  Human Brain Mapping (in press)
  [abstract] [last draft,soon]
  

  Predictive modeling of fMRI has the potential to expand the amount of information extracted and to enhance our understanding of brain systems by predicting brain states, rather than emphasizing the standard spatial mapping. Based on the block datasets of FIAC subject three, we demonstrate the potential and pitfalls of predictive modeling in fMRI analysis by investigating the performance of five models (linear discriminant analysis, logistic regression, linear support vector machine, Gaussian naïve Bayes and a variant) as a function of preprocessing steps and feature selection methods. We found that: (1) independent of the model, temporal detrending and feature selection assisted in building a more accurate predictive model; (2) the linear support vector machine and logistic regression often performed better than either of the Gaussian naïve Bayes models in terms of the optimal prediction accuracy; (3) the optimal prediction accuracy obtained in a feature space using principal components was typically lower than that obtained in a voxel space, given the same model and same preprocessing. We showed that due to the existence of artifacts from different sources, high prediction accuracy alone did not guarantee that the classifier was learning a pattern of brain activity that might be usefully visualized, though cross-validation methods do provide fairly unbiased estimates of true prediction accuracy. The trade-off between the prediction accuracy and the reproducibility of the spatial pattern should be carefully considered in predictive modeling of fMRI. We suggest that unless the experimental goal is brain-state classification of new scans on well-defined spatial features, prediction alone should not be used as an optimization procedure in fMRI data analysis.
• "Learning to Decode Cognitive States from Brain Images"
  
  Mitchell T., Hutchinson R., Niculescu S., Pereira F., Wang X., Just M., Newman S.
  Machine Learning Journal, Vol. 57, Issue 1-2, pp. 145-175, 2004
  [abstract] [paper]
  

  Over the past decade, functional Magnetic Resonance Imaging (fMRI) has emerged as a powerful new instrument to collect vast quantities of data about activity in the human brain. A typical fMRI experiment can produce a three-dimensional image related to the human subject's brain activity every half second, at a spatial resolution of a few millimeters. As in other modern empirical sciences, this new instrumentation has led to a flood of new data, and a corresponding need for new data analysis methods. We describe recent research applying machine learning methods to the problem of classifying the cognitive state of a human subject based on fRMI data observed over a single time interval. In particular, we present case studies in which we have successfully trained classifiers to distinguish cognitive states such as (1) whether the human subject is looking at a picture or a sentence, (2) whether the subject is reading an ambiguous or non-ambiguous sentence, (3) whether the word the subject is viewing is a noun or a verb, and (4) whether the noun the subject is viewing is a word describing food, people, buildings, etc. This learning problem provides an interesting case study of classifier learning from extremely high dimensional ($10^5$ features), extremely sparse (tens of training examples), noisy data. This paper summarizes the results obtained in these four case studies, as well as lessons learned about how to successfully apply machine learning methods to train classifiers in such settings.
• "Detecting Significant Multidimensional Spatial Clusters"
  
  Neill D., Moore A.,Pereira F., Mitchell T.
  Neural Information Processing Systems 2004
  
• "Classifying Instantaneous Cognitive States from fMRI Data"
  
  Mitchell T., Hutchinson R., Niculescu S., Pereira F., Wang X., Just M., Newman S.
  Proceedings of AMIA 2003 (Best Foundational Paper Award)
  
• "Distinguishing natural language processes on the basis of fMRI-measured brain activation"
  
  Pereira F., Just M. and Mitchell T.
  Principles of Knowledge Discovery in Databases 2001
  
• "Concise, Intelligible, and Approximate Profiling of Multiple Classes"
  
  Valdes-Perez R., Pericliev V. and Pereira F.
  International Journal of Human Computer Systems, 53(3):411-436, 2000
  

invited presentations

• "Challenges and limitations of classifiers for analysing fMRI data"
  
  as a coorganizer of the "New Directions on Decoding Mental States from fMRI Data" at NIPS 06 [slides]
• "Beyond Brain Blobs: machine learning classifiers as instruments for the analysis of fMRI data"
  
  this version encompasses talks at:
  Computer Science Department and Rotman Research Institute, University of Toronto
  Center for the Study of Brain, Mind and Behavior, Princeton University [slides]
• "Machine learning in the analysis of fMRI data: how to, what for, where next"
  
  Section on Functional Imaging Methods at NIH, [slides]
• "Analysis of Brain Imaging Data With Machine Learning Methods"
  
  Departamento de Neurologia y Neurocirugia, Universidad de Navarra, [slides]

software