WB-graphs: a within versus between group similarity interplay
(2022) , 17 pages
Files
ISBA_DP_2022-07.pdf
Open Access - Adobe PDF
- 634.47 KB
Details
- Authors
- Abstract
- A method for balancing the within/between group similarity is proposed in the frameworkof estimating graphs when data from multiple groups or classes are available. The method leverages connections between the Karush-Kuhn-Tucker (KKT) conditions for estimating ℓ1 penalized graphs in order to define an optimization problem that can be solved with already existing, known algorithms from the literature. The method is illustrated on an fMRI dataset and with a simulated, controlled experiment. Statistical guarantees are as well provided.
- Affiliations
Citations
Pircalabelu, E. (2022). WB-graphs: a within versus between group similarity interplay (LIDAM Discussion Paper ISBA 2022/07). https://hdl.handle.net/2078.5/110027