English

Continuous Learning in Single-Incremental-Task Scenarios

Machine Learning 2019-01-24 v3 Artificial Intelligence Computer Vision and Pattern Recognition Neural and Evolutionary Computing Machine Learning

Abstract

It was recently shown that architectural, regularization and rehearsal strategies can be used to train deep models sequentially on a number of disjoint tasks without forgetting previously acquired knowledge. However, these strategies are still unsatisfactory if the tasks are not disjoint but constitute a single incremental task (e.g., class-incremental learning). In this paper we point out the differences between multi-task and single-incremental-task scenarios and show that well-known approaches such as LWF, EWC and SI are not ideal for incremental task scenarios. A new approach, denoted as AR1, combining architectural and regularization strategies is then specifically proposed. AR1 overhead (in term of memory and computation) is very small thus making it suitable for online learning. When tested on CORe50 and iCIFAR-100, AR1 outperformed existing regularization strategies by a good margin.

Keywords

Cite

@article{arxiv.1806.08568,
  title  = {Continuous Learning in Single-Incremental-Task Scenarios},
  author = {Davide Maltoni and Vincenzo Lomonaco},
  journal= {arXiv preprint arXiv:1806.08568},
  year   = {2019}
}

Comments

26 pages, 13 figures; v3: major revision (e.g. added Sec. 4.4), several typos and minor mistakes corrected

R2 v1 2026-06-23T02:38:13.100Z