About this document
Unsupervised Outlier Detection Methods by Swetha Pattipaka is a document available to read on EtoBox.
The document discusses reverse nearest neighbors in unsupervised outlier detection. It notes that the commonly held view is that distance becomes meaningless in high dimensions due to concentration, making all points equal outliers. However, the paper aims to challenge this view by demonstrating distance-based methods can produce more contrasting outlier scores. It examines how some points appear infrequently in k-NN lists (antihubs), and the connection between antihubs, outliers, and outlier detection meth
- Author
- Swetha Pattipaka
- Language
- EN