Scalable adversarial online continual learning
Files
(Published version)
Date
2023
Authors
Dam, T.
Pratama, M.
Ferdaus, M.D.M.
Anavatti, S.
Abbas, H.
Editors
Amini, M.R.
Canu, S.
Fischer, A.
Guns, T.
Novak, P.K.
Tsoumakas, G.
Canu, S.
Fischer, A.
Guns, T.
Novak, P.K.
Tsoumakas, G.
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
Conference paper
Citation
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2023 / Amini, M.R., Canu, S., Fischer, A., Guns, T., Novak, P.K., Tsoumakas, G. (ed./s), vol.13715, pp.373-389
Statement of Responsibility
Conference Name
European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECMLPKDD) (19 Sep 2022 - 23 Sep 2022 : Grenoble, France)
Abstract
Adversarial continual learning is effective for continual learning problems because of the presence of feature alignment process generating task-invariant features having low susceptibility to the catastrophic forgetting problem. Nevertheless, the ACL method imposes considerable complexities because it relies on task-specific networks and discriminators. It also goes through an iterative training process which does not fit for online (one-epoch) continual learning problems.This paper proposes a scalable adversarial continual learning (SCALE) method putting forward a parameter generator transforming common features into task specific features and a single discriminator in the adversarial game to induce common features. The training process is carried out in meta-learning fashions using a new combination of three loss functions. SCALE outperforms prominent baselines with noticeable margins in both accuracy and execution time
School/Discipline
Dissertation Note
Provenance
Description
Access Status
Rights
Copyright 2022 The Author(s), under exclusive license to Springer Nature Switzerland
Access Condition Notes: Author supplied manuscript is available open access