Automatic Machine Learning with Auto^{n}ML with Piggy Yarroll


Details
Description
In the field of machine learning there are a lot of different algorithms with diverse properties. Understanding these properties can be a barrier to adoption. To address this issue, the Auton Lab at CMU has produced Auto^{n}ML, an automatic ML tool that recommends pipelines for a given problem.
Upon starting at the Auton Lab in the beginning of April 2023, Piggy was given the task of rearchitecting Auto^{n}ML to be more extensible. One key goal was to be able to add a new ML model in less than half an hour, a goal which we have now exceeded several times.
Piggy will talk about the challenges of building a system with very little prior knowledge of the field. He will cover the new architecture and how it captures the objective of providing a framework not just for ML models, but for other ML algorithms. The head of the Auton Lab would like every grad student to be able to add their work to Auto^{n}ML to make it available to data analysts inside the Auton Lab, and to less experienced data analysts out in the wild.
Piggy will show you how you can set up the open source Auto^{n}ML for your own projects.
La Monte H. P. "Piggy" Yarroll Bio
Piggy is a charter member of UniForum Chicago and one of the first sysadmins of the Clout Project. He has been programming for 40 years.
After an 11 year stint at Google, he has moved to Carnegie Mellon University to be a Senior Research Programmer in the Auton Lab. The Auton Lab is an applied research lab focused on Machine Learning.
La Monte has eclectic interests ranging from computers and electronics to abstract algebra, foreign languages, music, bushwalking, cheesemaking, and father of five :-).
Getting To The Meeting
This will be a virtual meeting on Zoom. The link to this Zoom meeting will be found in this associated Meetup.com announcement once you RSVP Yes under its Date/Time as its Location. See above under the meeting's title where it says the meetup.com link.
Zoom session will open at 6:45pm (US Central time - Chicago) with the meeting starting at 7:00pm ending around 9:00pm (with a short review of Group business before the speaker starts). There will be a meeting after the meeting for those who want to discuss other things. This meeting will be run virtually on-line using our UniForum Chicago Zoom account.
Note
In order to receive an email notification about meetings be sure email updates are turned on, and check your Meetup account at this location https://www.meetup.com/account/comm/ and there is a checkmark for new event announcements at this location https://www.meetup.com/account/comm/UniForum-Chicago/.
Our live meetings have been held previously at the Bolingbrook Fountaindale Library but we have been holding our meetings virtually since the pandemic began since March 2020. However, there is now the possibility of holding live meetings going forward but will be decided when and where live meetings will be held as they are defined going forward.
Please email info@uniforumchicago if you have a topic or speaker for a possible presentation at UniForum Chicago.
If you wish to be informed about future UniForum Chicago events you can sign up for our email list here.

Automatic Machine Learning with Auto^{n}ML with Piggy Yarroll