Together with @grfrederic we ported GMMs from a Snakemake workflow to bmi package, so that they can be used to provide model-based mutual information estimates.
This PR is accompanied with a subset of benchmark tasks with low-dimensional distributions on which we hypothesise the GMMs to obtain good performance.
Together with @grfrederic we ported GMMs from a Snakemake workflow to
bmipackage, so that they can be used to provide model-based mutual information estimates. This PR is accompanied with a subset of benchmark tasks with low-dimensional distributions on which we hypothesise the GMMs to obtain good performance.