molenc
molenc copied to clipboard
support weighted Faulon-*
Weights are > 0. The sum of weights could be 1. Try with Faulon-0,1,2. The weights could be optimized to maximize neighb. behavior.
Visualizing the different resulting MST might be insightful.
related to https://github.com/UnixJunkie/molenc/issues/41
related to L_1 and "fractional distance metrics" related to https://github.com/UnixJunkie/molenc/issues/82
this would be best optimized on a regression dataset
some regression benchmark tells me it does NOT outperform using a 2048b ECFP4...