Robust Classification of Dynamic Bichromatic point Sets in R2
Computational Geometry
2024-07-01 v2
Abstract
Let be a set of points in , and let . Our goal is to compute a line that "best" separates the "red" points from the "blue" points with at most outliers. We present an efficient semi-online dynamic data structure that can maintain whether such a separator exists. Furthermore, we present efficient exact and approximation algorithms that compute a linear separator that is guaranteed to misclassify at most , points and minimizes the distance to the farthest outlier. Our exact algorithm runs in time, and our -approximation algorithm runs in time. Based on our -approximation algorithm we then also obtain a semi-online data structure to maintain such a separator efficiently.
Keywords
Cite
@article{arxiv.2406.19161,
title = {Robust Classification of Dynamic Bichromatic point Sets in R2},
author = {Erwin Glazenburg and Frank Staals and Marc van Kreveld},
journal= {arXiv preprint arXiv:2406.19161},
year = {2024}
}
Comments
43 pages, 32 figures