Continual Learning by Three-Phase Consolidation
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910377900507136 |
|---|---|
| author | Maltoni, Davide Pellegrini, Lorenzo |
| author_facet | Maltoni, Davide Pellegrini, Lorenzo |
| contents | TPC (Three-Phase Consolidation) is here introduced as a simple but effective approach to continually learn new classes (and/or instances of known classes) while controlling forgetting of previous knowledge. Each experience (a.k.a. task) is learned in three phases characterized by different rules and learning dynamics, aimed at removing the class-bias problem (due to class unbalancing) and limiting gradient-based corrections to prevent forgetting of underrepresented classes. Several experiments on complex datasets demonstrate its accuracy and efficiency advantages over competitive existing approaches. The algorithm and all the results presented in this paper are fully reproducible thanks to its publication on the Avalanche open framework for continual learning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_14679 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Continual Learning by Three-Phase Consolidation Maltoni, Davide Pellegrini, Lorenzo Machine Learning Computer Vision and Pattern Recognition TPC (Three-Phase Consolidation) is here introduced as a simple but effective approach to continually learn new classes (and/or instances of known classes) while controlling forgetting of previous knowledge. Each experience (a.k.a. task) is learned in three phases characterized by different rules and learning dynamics, aimed at removing the class-bias problem (due to class unbalancing) and limiting gradient-based corrections to prevent forgetting of underrepresented classes. Several experiments on complex datasets demonstrate its accuracy and efficiency advantages over competitive existing approaches. The algorithm and all the results presented in this paper are fully reproducible thanks to its publication on the Avalanche open framework for continual learning. |
| title | Continual Learning by Three-Phase Consolidation |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2403.14679 |