Selective Attention-based Modulation for Continual Learning
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_ | 1866916184134254592 |
|---|---|
| author | Bellitto, Giovanni Salanitri, Federica Proietto Pennisi, Matteo Boschini, Matteo Porrello, Angelo Calderara, Simone Palazzo, Simone Spampinato, Concetto |
| author_facet | Bellitto, Giovanni Salanitri, Federica Proietto Pennisi, Matteo Boschini, Matteo Porrello, Angelo Calderara, Simone Palazzo, Simone Spampinato, Concetto |
| contents | We present SAM, a biologically-plausible selective attention-driven modulation approach to enhance classification models in a continual learning setting. Inspired by neurophysiological evidence that the primary visual cortex does not contribute to object manifold untangling for categorization and that primordial attention biases are still embedded in the modern brain, we propose to employ auxiliary saliency prediction features as a modulation signal to drive and stabilize the learning of a sequence of non-i.i.d. classification tasks. Experimental results confirm that SAM effectively enhances the performance (in some cases up to about twenty percent points) of state-of-the-art continual learning methods, both in class-incremental and task-incremental settings. Moreover, we show that attention-based modulation successfully encourages the learning of features that are more robust to the presence of spurious features and to adversarial attacks than baseline methods. Code is available at: https://github.com/perceivelab/SAM. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_20086 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Selective Attention-based Modulation for Continual Learning Bellitto, Giovanni Salanitri, Federica Proietto Pennisi, Matteo Boschini, Matteo Porrello, Angelo Calderara, Simone Palazzo, Simone Spampinato, Concetto Computer Vision and Pattern Recognition We present SAM, a biologically-plausible selective attention-driven modulation approach to enhance classification models in a continual learning setting. Inspired by neurophysiological evidence that the primary visual cortex does not contribute to object manifold untangling for categorization and that primordial attention biases are still embedded in the modern brain, we propose to employ auxiliary saliency prediction features as a modulation signal to drive and stabilize the learning of a sequence of non-i.i.d. classification tasks. Experimental results confirm that SAM effectively enhances the performance (in some cases up to about twenty percent points) of state-of-the-art continual learning methods, both in class-incremental and task-incremental settings. Moreover, we show that attention-based modulation successfully encourages the learning of features that are more robust to the presence of spurious features and to adversarial attacks than baseline methods. Code is available at: https://github.com/perceivelab/SAM. |
| title | Selective Attention-based Modulation for Continual Learning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2403.20086 |