Generalization Bounds for Data-Driven Numerical Linear Algebra

Data-driven algorithms can adapt their internal structure or parameters to inputs from unknown application-specific distributions, by learning from a training sample of inputs. Several recent works have applied this approach to problems in numerical linear algebra, obtaining significant empirical ga...

Full description

Bibliographic Details

Main Authors:	Bartlett, Peter, Indyk, Piotr, Wagner, Tal
Format:	Article
Published:	2022
Subjects:	Machine Learning Data Structures and Algorithms
Online Access:	https://arxiv.org/abs/2206.07886 http://arxiv.org/abs/2206.07886
Tags:	Add Tag No Tags, Be the first to tag this record!