AI/ML Course Week 5

AI/ML Course Week 5: In lieu of dedicated ABCD-ReproNim instructors for this week, we will link to several relevant (and excellent!) lectures and resources on explainable machine learning and FAIRness in machine learning.

Help and Resources: If you have questions about this week's material please post them to the ABCD-ReproNim Neurostars or the ABCD-ReproNim Slack.

AI/ML Course Week 5 Assignments:

AI/ML Week 5 Data Exercise (Jupyter notebook version)

AI/ML Week 5 Data Exercise Submission Form

Project Month Prep 3: hierarchical clustering of features from ABCD 4.0 (this assignment is optional and, you won't submit anything for it, but if you plan to attend Project Month we recommend trying it!)

Announcements:

AI/ML Course links: Check out our AI/ML Course syllabus. If you haven’t already, please make sure to register so we can add you to our AI/ML Course email list. Here is an archive of information emails from previous weeks of the course. AI/ML Course materials will be updated weekly and will remain online indefinitely.

AI/ML Week 5 Data Exercise: This week’s exercise is required for all students wishing to receive an ABCD-ReproNim AI/ML Course Certificate of Completion and/or an invitation to Project Month. Please complete the AI/ML Week 5 Data Exercise after viewing the Week 5 videos. Please push your code to GitHub and submit a link to your file here. We do not “grade” your submissions for accuracy. Rather, we use them to determine who is participating in the course and can receive a Certificate of Completion/invitation to Project Month (April 4 - 29, 2022). Solutions for this week’s data exercise can be found on here.

Project Month Registration: The ABCD-ReproNim Project Month (April 4-29, 2022) is next week! If you would like to take part in Project Month please fill out our Project Month Registration Form. Here is information on the planned schedule of events and what to expect. ABCD-ReproNim students are invited to participate in project Month if they have an active ABCD Data Access Confirmation (DUC) and have completed at least 80% of the AI/ML Course weekly Data Exercises including the Week 5 exercise on FAIRness in ML (or 70% or the Main Course Data Exercises).

ABCD Data Access Confirmation: If you wish to access ABCD data you will need an active Data Use Certification (DUC) from the NDA. We provide instructions on how to obtain ABCD data access to help with this process. Once you have an approved DUC please complete our ABCD Data Access Confirmation. Note that you do not need to have ABCD data access in order to take the AI/ML Course. This assignment is optional and does not count towards completion of the course. However, students who do not successfully complete this assignment will not be allowed to participate in Project Month.

ABCD-ReproNim JupyterHub Access Request: ABCD-ReproNim students with active DUCs can access to the new ABCD-ReproNim JupyterHub, a cloud-based scientific computing platform that is fully accessible through an internet browser (see this link for details about the JupyterHub project!) This document introduces the ABCD-ReproNim JupyterHub, how to access it, and outlines some basic usage rules. Students can use the Hub to work on Data Exercises and perform ABCD data analysis projects during the course. ABCD-ReproNim staff will grant access to students with active DUCs who submit an Access Request. Students without an active NDA Data Use Certification are invited to follow the instructions here where they can access the same computational environment locally via Docker.