Machine Learning for the Cognitive Neurosciences


                                       



From the 18th to 22nd of January, 2021, the NeuroDataScience ORIGAMI lab, the McGill Centre for Integrative Neuroscience, HBHL’s NeuroHub, ReproNim, and UNIQUE will hold a workshop entitled “Machine Learning for the Cognitive Neurosciences”, whose objective will be to introduce researchers to both the conceptual and practical rudiments of machine learning as applied to neuroimaging. See the schedule below.

Though attendees are expected to have a general familiarity with brain imaging and conventional statistical analyses, the first two days introduce attendees to the software environment that will serve for the practical exercises during the workshop, including some basics of the Linux command line (Bash), Git, the Python programming language, and several Python-based packages for scientific computing, courtesy of Calcul Québec.

The core content will be presented over the remaining days. Our first speaker, Dr. Tal Yarkoni (University of Texas at Austin) will discuss the scientific applications of machine learning in brain imaging. Our second speaker, Dr. Ella Gabitov (McGill University), will present on the explanatory power and limitations of machine learning when applied to cognitive neuroscience. Throughout the workshop, hands-on tutorials will be presented by Alexandre Hutton (ORIGAMI, McGill), which will focus on giving participants an understanding of core concepts and common practices in machine learning.

Logistic Summary:

Dates: January 18-22, 2021
Cost: $10
Venue: The workshop will be held online over Zoom.
Registration: Closed.
Contact: info.mlcn@gmail.com
Material: Link!
YouTube streaming: Link!

Volunteers

Want to help out? Contact us at info.mlcn@gmail.com, we’re looking for people to help out with answering questions and solving users’ technical issues.
Here are the people who have generously offered to help out:
Nikhil Bhagwat
Peer Herholz
Vincent Qing
Jérôme Dockès
Kendra Oudyk
Jacob Sanz-Robinson

Organizers & Contributors:

Alexandre Hutton, MNI, McGill University
Paule-Joanne Toussaint, MNI, McGill University
Patrick Bermudez, MNI, McGill University
Alan C. Evans, MCIN, McGill University
Jean-Baptiste Poline, MNI, McGill University

NeuroDataScience: ORIGAMI
McGill Centre for Integrative Neuroscience
HBHL-NeuroHub
UNIQUE

Schedule:

Groups:
To better support participants in the initial days of the workshop, the groups are split into two. If you registered for the Python intro on Monday, you are in Group 1. Otherwise, you are in Group 2.

To set up your environment, follow these instructions; we recommend that you do this ahead of time (est. 15min). Contact us via Slack if you’d like some help (check your email for the invitation).

Day 1 (Monday)
Group 1:
09:00 – 12:00 Calcul Québec – Introduction to Programming with Python
Group 2:
09:00 - 10:00 Peer Herholz - Linux command line: Bash.
10:00 - 12:00 Kendra Oudyk - Git & version control

Day 2 (Tuesday)
Group 1:
09:00 - 10:00 Peer Herholz - Linux command line: Bash.
10:00 - 12:00 Kendra Oudyk - Git & version control
Group 2:
09:00 – 12:00 Calcul Québec – Introduction to Programming with Python

Day 3 (Wednesday)
09:00 – 10:30 Dr. Tal Yarkoni – “Machine Learning as Applied to Neuroimaging”
10:30 – 12:00 Alexandre Hutton – Hands-on ML: dataset handling, sklearn/nilearn, estimators & models, model evaluation
13:30 – 16:00 Practical exercises (offline)

Day 4 (Thursday)
09:00 – 12:00 Alexandre Hutton – Hands-on ML: cross-validation, ROC & AUC, visualization, evaluation metrics
13:30 – 16:00 Practical exercises (offline)

Day 5 (Friday)
09:00 – 10:00 Dr. Ella Gabitov, Dr. Jean-Baptiste Poline – “The elephant in the room: to explain or to predict, that is the question”
10:00 – 12:00 Alexandre Hutton – Hands-on ML: confounds. model interpretation (and lack thereof)
13:30 – 16:30 Practical exercises (offline), data clinic