Local determination of nearest-nbrs: MI/CMI forest estimators with G-KSG (geodesic distance using Unsupervised forest) and Local Permutation

[SUKI-O]

trafficstars

Is your feature request related to a problem? Please describe. Would like to get a Mutual Information / Conditional MI using geodesic distance derived from Unusupervised Forest as described in the following papers:

• Estimating Mutual Information
• Estimating Mutual Information via Geodesic kNN

Describe the solution you'd like

• experiment with the proposed algorithm(s) leveraging existing features in sktree and additional Python code
• add/change Cython module(s) that will be needed for implementation
• add tests to verify the results and performance

Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered.

Additional context Add any other context or screenshots about the feature request here.

[icy-23]

I'm interested in this, I also would be interested in general KL Divergence estimation

[adam2392]

Reelvant too: https://projecteuclid.org/journals/annals-of-statistics/volume-50/issue-6/Local-permu[…]-tests-for-conditional-independence/10.1214/22-AOS2233.short