Quadratic Suffices for Over-parametrization via Matrix Chernoff Bound

06/09/2019

∙

We improve the over-parametrization size over two beautiful results [Li and Liang' 2018] and [Du, Zhai, Poczos and Singh' 2019] in deep learning theory.

READ FULL TEXT

Quadratic Suffices for Over-parametrization via Matrix Chernoff Bound

Computing the Quadratic Numerical Range

How Could Polyhedral Theory Harness Deep Learning?

Codebooks from generalized bent Z_4-valued quadratic forms

An Improved Bound for the Nystrom Method for Large Eigengap

A structure-preserving doubling algorithm for solving a class of quadratic matrix equation with M-matrix

Optimal quadratic binding for relational reasoning in vector symbolic neural architectures

Deep Graph Matching under Quadratic Constraint

Quadratic Suffices for Over-parametrization via Matrix Chernoff Bound

Related Research

Computing the Quadratic Numerical Range

How Could Polyhedral Theory Harness Deep Learning?

Codebooks from generalized bent Z_4-valued quadratic forms

An Improved Bound for the Nystrom Method for Large Eigengap

A structure-preserving doubling algorithm for solving a class of quadratic matrix equation with M-matrix

Optimal quadratic binding for relational reasoning in vector symbolic neural architectures

Deep Graph Matching under Quadratic Constraint