# CoSMo 2013 Multivariate Pattern Analysis Workshop¶

These exercises were presented at the CoSMo 2013 workshop in Kingston, Ontario, Canada.

History of exercises:

- 2019 NMSM 2019, Noesselt’s lab 3rd Modelling Symposium, University of Magdeburg.
- 2017 LABMAN 2017, the second Latin American Brain Mapping Network Meeting.
- 2016 PRNI 2016, the 6th International Workshop on Pattern Recognition in Neuroimaging workshop.
- 2016 CIMeC 2016 Hands-on Methods.
- 2016 Royal Holloway University of London CoSMoMVPA workshop.
- 2014 CIMeC 2014 Advanced Neural Decoding.
- 2013 2013 Summer School in Computational Sensory-Motor Neuroscience.

## Contents¶

- Introduction
- Exercises
- Dataset basics
- Split-half correlation-based MVPA with group analysis
- Classification with Naive Bayes classifier
- Write your own Nearest-neighbor classifier
- Build wrapper for Matlab’s SVM classifier
- Cross-validation part 1: N-Fold Partitioner
- Cross-validation part 2: using multiple classifiers
- Cross-validation part 3: using a dataset measure
- Visualization of DSMs
- Representational similarity analysis
- Define voxel selection for a searchlight
- Write a searchlight function that computes a generic dataset measure
- Use the searchlight with a neighborhood and a measure

Matlab outputs: matlab_pb_toc