Lifelong Representation Learning


Given a set of problems to solve, the dominant paradigm in the AI community has been to solve each problem or task independently. This is in sharp contrast with the human capability to build from past experience and transfer knowledge to speed-up the learning process for a new task. To mimic such a capability, the machine learning community has introduced the concept of continual learning or lifelong learning. The main advantage of this paradigm is that it enables learning with less data, it often allows to learn faster and to generalize better. From an industrial standpoint, the potential of lifelong learning is tremendous as this would mean deploying machine learning models faster by bypassing the need to collect labels.


Our chair is structured around several lines of research including self-supervised learning, continual adaptation, online learning, and stream learning.

We collaborate with the chair lead by Julien Mairal, entitled Towards More Data Efficiency in Machine Learning.

Chair events

Selected list of publications

  • On the Road to Online Adaptation for Semantic Image Segmentation. R. Volpi, P. De Jorge, D. Larlus, G. Csurka. Computer Vision and Pattern Recognition conference (CVPR), 2022.
  • Concept Generalization in Visual Representation Learning. M. B. Sariyildiz, Y. Kalantidis, D. Larlus, K. Alahari. (ICCV), 2021.
  • Continual Adaptation of Visual Representations via Domain Randomization and Meta-learning. R. Volpi, D. Larlus, G. Rogez. Computer Vision and Pattern Recognition conference (CVPR), 2021.
  • Hard Negative Mixing for Contrastive Learning. Y. Kalantidis, M. B. Sariyildiz, N. Pion, P. Weinzaepfel, D. Larlus. (NeurIPS) 2020.
  • Learning Visual Representations with Caption Annotations. B. Sariyildiz, J. Perez, D. Larlus. European Conference on Computer Vision (ECCV) 2020.