Scalable adversarial online continual learning

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.

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

License

Grant ID

Call number

Persistent link to this record