Special Session at the 2022 IEEE International Conference on Evolving and Adaptive Intelligent Systems (EAIS22)
Location : Larnaca, Cyprus
Date : May 25-27, 2022
Continual learning requires the ability to quickly adapt to novel data and exploit existing information without forgetting previous knowledge. As such, the performance of a continual learning agent needs to be measured in different learning environments through multiple metrics (forward / backward transfer, computational complexity, sample efficiency…).
However, current continual learning literature mostly focuses on the mitigation of catastrophic forgetting in a restricted set of tasks and learning environments, such as object recognition in computer vision.
In this special session, we aim at expanding the current set of continual learning scenarios, metrics and strategies with new proposals in order to better address the objectives of the continual learning challenge. We particularly welcome broad experimental evaluations taking into account heterogeneous scenarios and metrics, design of new strategies which do not focus solely on catastrophic forgetting and assessment of existing continual learning approaches in real-world applications.
The topics of interest for the special session include, but are not limited to:
continual learning scenarios and learning environments
continual learning benchmarks
bio-inspired continual learning strategies
evaluation protocols and metrics for continual learning
continual learning for sequential data processing tasks
meta continual learning and continual meta learning
task-free and online continual learning
continual reinforcement learning
real-world impact and applications of continual learning
continual learning for federated environments
continual learning on the edge
main conference webpage: http://cyprusconferences.org/eais2022/
Paper submission: January 10, 2022 February 7, 2022
Notification of acceptance/rejection: February 19, 2022
Camera ready submission: March 20, 2022
Authors registration: March 20, 2022
Conference Dates: May 25-27, 2022
Submitted papers should not exceed 8 pages plus at most 2 pages overlength.
Submissions of full papers are accepted online through the EasyChair system.
Please check the main conference website for more details.
Antonio Carta (University of Pisa, Italy), Andrea Cossu (University of Pisa, Italy), Vincenzo Lomonaco (University of Pisa, Italy), Davide Maltoni (University of Bologna, Italy)