Abstract
This study investigates a robust measure of similarity applicable in many domains and across many dimensions of data. Given a distance or discrepancy measure on a domain, the similarity of two values in this domain is defined as the probability that any pair of values from that domain are more different (at a larger distance) than these two values are. We discuss the motivation for this approach, its properties, and the issues that arise from it. © 2010 Springer-Verlag Berlin Heidelberg.
Cite
CITATION STYLE
Ralescu, A., Visa, S., & Popovici, S. (2010). A stochastic treatment of similarity. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6178 LNAI, pp. 11–18). https://doi.org/10.1007/978-3-642-14049-5_2
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
