2021-03-18-doan21a.md

title	abstract	layout	series	publisher	issn	id	month	tex_title	firstpage	lastpage	page	order	cycles	bibtex_author	author	date	address	container-title	volume	genre	issued	pdf	extras
A Theoretical Analysis of Catastrophic Forgetting through the NTK Overlap Matrix	Continual learning (CL) is a setting in which an agent has to learn from an incoming stream of data during its entire lifetime. Although major advances have been made in the field, one recurring problem which remains unsolved is that of Catastrophic Forgetting (CF). While the issue has been extensively studied empirically, little attention has been paid from a theoretical angle. In this paper, we show that the impact of CF increases as two tasks increasingly align. We introduce a measure of task similarity called the NTK overlap matrix which is at the core of CF. We analyze common projected gradient algorithms and demonstrate how they mitigate forgetting. Then, we propose a variant of Orthogonal Gradient Descent (OGD) which leverages structure of the data through Principal Component Analysis (PCA). Experiments support our theoretical findings and show how our method can help reduce CF on classical CL datasets.	inproceedings	Proceedings of Machine Learning Research	PMLR	2640-3498	doan21a	0	A Theoretical Analysis of Catastrophic Forgetting through the NTK Overlap Matrix	1072	1080	1072-1080	1072	false	Doan, Thang and Abbana Bennani, Mehdi and Mazoure, Bogdan and Rabusseau, Guillaume and Alquier, Pierre	given family Thang Doan given family Mehdi Abbana Bennani given family Bogdan Mazoure given family Guillaume Rabusseau given family Pierre Alquier	2021-03-18		Proceedings of The 24th International Conference on Artificial Intelligence and Statistics	130	inproceedings	date-parts 2021 3 18	http://proceedings.mlr.press/v130/doan21a/doan21a.pdf	label link Supplementary ZIP http://proceedings.mlr.press/v130/doan21a/doan21a-supp.zip