Machine learning for neuroimaging …

… with Scikit-learn and nilearn

Team: Pierre Bellec, Elizabeth DuPre, Greg Kiar, Jacob Vogel

Date: December 12th, 9h-17h. Breakfast/registration at 8h30.

Location: Amphithéâtre “le groupe Maurice”, CRIUGM

Summary: This course will be a hands-on/type-along introduction to machine learning for neuroimaging problems with scikit-learn and nilearn.

Morning (9h-12h30): introduction to machine-learning with scikit-learn

This part of the course will follow the scikit-learn chapter of the scipy-lectures, found here. This includes:

• Basic principles
• Supervised learning: classification, the example of handwritten digits
• Supervised learning: regression, the example of housing data
• Measuring prediction performance
• Unsupervised learning: dimension reduction and visualization
• Chaining estimators: the example of eigenfaces
• Parameter selection, validation, and testing

Afternoon (13h30-17h): introduction to nilearn

This part of the course will provide a general introduction to nilearn, building off of several example analyses.

Prerequisites

• Basic familiarity with Python would be preferable
• You will need enough space for Anaconda and all the course data (~4GB).

If you are already savvy with Python and just want a tl;dr summary, here’s all you need to know:

1. Join the Brainhack Slack group and join the main-nilearn-2018 channel
2. Download and install python with the full-suite 64-bit Anaconda distribution
3. Download the data and remember where you store it!
4. Download or clone the Intro to ML repository
5. Install the necessary packages: pip install -U nilearn scipy matplotlib scikit-learn jupyter pandas seaborn
6. Test everything by opening one of the MAIN-tutorial .ipynb notebooks and running the first few cells

For detailed instructions, view the full installation instructions.