Reproducible Research in Pattern Recognition by Unknown

Author:Unknown
Language: eng
Format: epub
ISBN: 9783030239879
Publisher: Springer International Publishing

2.1 Metric Estimation

The first step in computing ranking metrics of an embedding is to compute the distance matrix between any query and a database of size K for a specified distance metric, or equivalently a similarity matrix. The next step is to compute a relevance matrix with elements:

(1)

where y(x) denotes the class of sample x.

The row-wise sorting of R by the values in D, results in the so called correct matrix , containing elements that are 1 if the k-th closest sample of the database to the query q is relevant and 0 otherwise. The matrix C can be directly used to compute the precision and recall for any query and rank k resulting in the matrix Pr and Rc, respectively with elements:

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.