BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//YourSite//ICS Event Generator//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:event-1780540484558-99hdqepv@yourdomain.com
DTSTAMP:20260604T023444Z
DTSTART;TZID=Europe/Berlin:20260610T100000
DTEND;TZID=Europe/Berlin:20260610T110000
SUMMARY:CoLLAs Seminar: Rahaf Aljundi on Continuous Accumulation of Knowledge and Experience
DESCRIPTION:Dear all\,\n\nWe are excited to bring you the second speaker of the monthly CoLLAs Seminar Series\, which has an excellent line-up of speakers from around the world: https://lifelong-ml.cc/seminar \n\nWith this series\, we aim to bring together researchers working on continual\, lifelong\, and adaptive machine learning to share new ideas\, and foster community-wide dialogue!\n\nOur second talk will be by Rahaf Aljundi on June 10 at 10:00 AM CET.\n\nTitle: Towards Continuous Accumulation of Knowledge and Experience\n\nAbstract: Continual learning enables models to adapt to streaming data\, targeting knowledge accumulation\, and avoiding catastrophic forgetting. Continual learning research has focused on continual parameter updates\; however\, forgetting and learning instability always seem unavoidable\, rendering continual learning a far‑fetched problem. On the other hand\, in‑context learning offers a complementary path to frequent parameter updates. In this talk\, I argue that efficient\, continuous adaptation needs not occur solely in parameter space. I will show how memory‑based mechanisms enable rapid adaptation and present some of my previous and recent continual learning works\, then discuss how combining fast\, memory‑driven updates with slow model consolidation could shape the future of continual learning.\n\nSpeaker bio: Rahaf Aljundi is a research scientist at Toyota Motor Europe. She is interested in building models that can keep learning (given any source of supervision)\, can tell when a novel input is provided and further point at instances that require annotation. This includes the topics of continual learning (incremental learning\, lifelong learning)\, novelty detection (out of distribution\, anomaly detection)\, active learning and also domain adaptation. She has obtained her PhD degree from KU Leuven university (Belgium) where she extensively worked on the topic of continual learning.\n\nZoom: https://polymtl-ca.zoom.us/j/84066718835?pwd=2cpVMN2XjRXyZrngOFBJSO3Fwifywo.1
END:VEVENT
END:VCALENDAR