Covariance and Correlation Kernels on a Graph in the Generalized Bag-of-Paths Formalism
(2019) , 44 pages
Files
2019-03.pdf
Closed Access - Adobe PDF
- 505.81 KB
Details
- Authors
Courtain, Sylvain 
- Abstract
- This paper aims at deriving closed-form expressions computing the expectation of co-presences and of number of co-occurrences of nodes on paths sampled from a network according to general path weights (a bag of paths). The idea is that two nodes, when they appear together on (preferably short) paths of the network, are considered similar. The results are provided for both regular and hitting paths and are then used for computing new covariance and correlation measures between nodes. Semi-supervised classification experiments show that the introduced similarity measures provide competitive results compared to other state-of-the-art distances and similarities.
- Affiliations
Citations
Guex, G., Courtain, S., & Saerens, M. (2019). Covariance and Correlation Kernels on a Graph in the Generalized Bag-of-Paths Formalism (Louvain Research Institute in Management and Organizations Working Paper Series 2019/03). https://hdl.handle.net/2078.5/170803