Current browse context:
stat
Change to browse by:
References & Citations
Computer Science > Machine Learning
Title: Coresets via Bilevel Optimization for Continual Learning and Streaming
(Submitted on 6 Jun 2020 (v1), last revised 22 Oct 2020 (this version, v2))
Abstract: Coresets are small data summaries that are sufficient for model training. They can be maintained online, enabling efficient handling of large data streams under resource constraints. However, existing constructions are limited to simple models such as k-means and logistic regression. In this work, we propose a novel coreset construction via cardinality-constrained bilevel optimization. We show how our framework can efficiently generate coresets for deep neural networks, and demonstrate its empirical benefits in continual learning and in streaming settings.
Submission history
From: Zalán Borsos [view email][v1] Sat, 6 Jun 2020 14:20:25 GMT (1834kb,D)
[v2] Thu, 22 Oct 2020 17:53:39 GMT (3397kb,D)
Link back to: arXiv, form interface, contact.