Anomaly Detection in High Dimensional Data (Q74767)
From MaRDI portal
scientific article from arXiv
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Anomaly Detection in High Dimensional Data |
scientific article from arXiv |
Statements
12 August 2019
0 references
stat.ML
0 references
cs.LG
0 references
stat.AP
0 references
This article presents a novel approach to detecting anomalies in high-dimensional data, dubbed the "stray algorithm." The authors address shortcomings of the existing HDoutliers algorithm by proposing an innovative method that leverages extreme value theory to enhance threshold calculation efficiency. Extensive tests with both synthetic and actual datasets showcase the stray algorithm's superiority over HDoutliers in terms of accuracy and speed. To facilitate wider use, the stray algorithm is packaged as an open-source R package. (English)
0 references
Hey little buddy! Imagine you have a big box filled with colorful marbles, but one or two are different from the others. We need to find out which ones they are so we can separate them. In this story, people made a special tool called "the stray algorithm" that helps us see if any of our special marbles (called anomalies) stick out and need our attention. They compared their new method with an old one called HDoutliers, and guess what? Their new stray algorithm works better because it can find the different marbles faster and more accurately! Plus, they shared this cool tool for everyone to use by making it a free R package that anyone can play with. (English)
0 references
