Workshop on Hyperparameter Optimization and Distributed Training

AIDA invites our technical community to an AIDA-day on Zoom where we will deep dive into hot topics in AI training. Our keynote speaker, Veronika Cheplygina, IT University Copenhagen DK, is invited to present her recent paper on methodological failures in ML for medical imaging and to discuss her interesting work on tackling limited access to labelled data. We continue with a workshop structured to cover two important aspects of training deep neural networks. The workshop tutor Erik Ylipää, ENCCS/RISE, will guide us through the two sessions: “Hyperparameter optimization using Optuna” and “Distributed PyTorch training: single and multiple node training”. The workshop will mostly follow a “code-along” structure with material distributed in advance. You are welcome to prepare questions, both general and specific in detail, to discuss, break down and hopefully problem-solve. For those of you who are not technical experts we recommend you take part in AIDA’s regular internal meeting and listen to the keynote presentation. Further reads: https://www.nature.com/articles/s41746-022-00592-y – Veronika Cheplygina

Date and time: October 26, 13.00-17.00 (Zoom).

13.00-13.45 Keynote Veronika Cheplygina 13.45-17.00 Technical workshop

Please register (for the whole or a certain part of the AIDA day at):https://forms.office.com/r/emkTX7BsFA

Zoom link will be distributed to registered participants prior to the meeting Most Welcome! AIDA organizes these AIDA-day meetings for affiliated parties. As a part external to AIDA you are very welcome to visit us as a guest for 1-2 events to get a better picture of our organization. Please contact caroline.bivik.stadler@liu.se if you want more information about AIDA and the benefits of becoming a network partner.