Quadrature-based features for kernel approximation

Marina Munkhoeva, Yermek Kapushev, Evgeny Burnaev, Ivan Oseledets
2/11/2018
cs.LGstat.ML

Official code: maremun/quffka (16 stars, MIT, last commit 2018-10-30). 1 more implementation below.

Abstract

We consider the problem of improving kernel approximation via randomized feature maps. These maps arise as Monte Carlo approximation to integral representations of kernel functions and scale up kernel methods for larger datasets. Based on an efficient numerical integration technique, we propose a unifying approach that reinterprets the previous random features methods and extends to better estimates of the kernel approximation. We derive the convergence behaviour and conduct an extensive empirical study that supports our hypothesis.

View on arXivView PDF

Code Implementations(2)

maremun/quffkaOfficial100%
MIT

Quadrature-based features for kernel approximation

162Shell, PythonNov 20, 20177 years agoMIT
kernel-approximationkernel-methodsmachine-learningquadrature-rulesrandom-features
quffka/quffkaOfficial100%
MIT

Quadrature-based features for kernel approximation

41Shell, PythonApr 17, 20188 years agoMIT

Similar Papers

Revisiting Random Binning Features: Fast Convergence and Strong Parallelizability

84% match2 repos32018

Universal approximation property of Banach space-valued random feature models including random neural networks

83% match1 repo12023

Feature maps for the Laplacian kernel and its generalizations

83% match1 repo102025

On the Approximation of Kernel functions

82% match2024

Random feature approximation for general spectral methods

82% match2023

Randomized Quasi-Monte Carlo Features for Kernel Approximation

82% match2025

Cite this paper

@article{munkhoeva2018quadraturebased,
  title  = {Quadrature-based features for kernel approximation},
  author = {Marina Munkhoeva and Yermek Kapushev and Evgeny Burnaev and Ivan Oseledets},
  year   = {2018},
  eprint = {1802.03832},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1802.03832}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1802.03832.svg)](https://codeofpaper.com/paper/1802.03832)

Discussion