Categorical and dimensional variable prediction from resting state fMRI data, a new example and tutorial for NiLearn
The project we propose for Brainhack intends to build a tutorial on machine learning methods for rsfMRI data. As a minimum, we will use two different machine learning methods for classification and regression, Gaussian Naive Bayes and Support Vector Machine. We will focus on implementations that allow people to easily change and compare these different algorithms, using algorithms available at scikit-learn (scikit-learn.org). We will use the ABIDE dataset, which contain rsfMRI data of patients with autism spectrum disorder. For the classification demo, we propose to use data from all subjects in order to differentiate the three types autism of patients and also control subjects. For the regression demo, we expect to predict phenotipic variables available from this dataset. Our goal is for that project participants learn how to use these techniques, as well as to provide an educational resource for others since we expect to incorporate this tutorial in NiLearn.
Initially I can teach participants the benefits of scikit-learn and how to configure sci-kit learn and also run tutorials available on their website (http://nilearn.github.io/). I can also help in issues regarding machine learning algorithms and python.
contact Caroline Froehlich (email@example.com) and Alexandre Franco (firstname.lastname@example.org)
Caroline Froehlich Nathassia Aurich Alexandre Franco