11:00
12:00

GT AlgoDist, «(Online) Continual Learning», Akka Zemari

Akka Zemari (LaBRI)

Title: (Online) Continual Learning

Abstract:

In real-world supervised learning, training data is often unavailable simultaneously, requiring models to adapt to incoming information. Sequential or naive training of pre-trained models on new tasks can lead to «forgetting» of prior knowledge. Incremental learning methods aim to adapt models to new data while retaining past knowledge. Focusing on the streaming scenario, where data arrives one sample at a time, our «Move-to-Data» method selectively adjusts network weights without systematic gradient descent. Compared to the state-of-the-art methods, our approach outperforms and learns significantly faster, presenting a promising solution for efficient and effective continual learning.

N.B. : No prerequisites are required to attend the presentation; we will provide the necessary reminders to make the presentation self-contained.

https://algodist.labri.fr/index.php/Main/GT

LaBRI salle 178 - lien visio https://webconf.u-bordeaux.fr/b/arn-4tr-7gp