latentmi.ksg

Functions

add_noise(x[, intens])
query_neighbors(tree, x, k)
count_neighbors(tree, x, r)
avgdigamma(points, dvec)	tweaked this to return list of psi(nx) for pointwise estimate
build_tree(points)
lnc_correction(tree, points, k, alpha)	in my experience this doesn't work in practice for N_dims > 1
entropy(x[, k, base])	The classic K-L k-nearest neighbor continuous entropy estimator
mi(x, y[, z, k, base, alpha])	Mutual information of x and y (conditioned on z if z is not None)
entropyd(sx[, base])	Discrete entropy estimator
centropyd(x, y[, base])	The classic K-L k-nearest neighbor continuous entropy estimator for the
midd(x, y[, base])	Discrete mutual information estimator
out_of_sample_avgdigamma(obs_points, est_points, dvec)	here we compute psi(nx) for query points that are not
out_of_sample_pmi(obs_x, obs_y, est_x, est_y[, k, base])

Module Contents

latentmi.ksg.add_noise(x, intens=1e-10)

latentmi.ksg.query_neighbors(tree, x, k)

latentmi.ksg.count_neighbors(tree, x, r)

latentmi.ksg.avgdigamma(points, dvec)

tweaked this to return list of psi(nx) for pointwise estimate

latentmi.ksg.build_tree(points)

latentmi.ksg.lnc_correction(tree, points, k, alpha)

in my experience this doesn’t work in practice for N_dims > 1

latentmi.ksg.entropy(x, k=3, base=2)

The classic K-L k-nearest neighbor continuous entropy estimator x should be a list of vectors, e.g. x = [[1.3], [3.7], [5.1], [2.4]] if x is a one-dimensional scalar and we have four samples

latentmi.ksg.mi(x, y, z=None, k=3, base=2, alpha=0)

Mutual information of x and y (conditioned on z if z is not None) x, y should be a list of vectors, e.g. x = [[1.3], [3.7], [5.1], [2.4]] if x is a one-dimensional scalar and we have four samples

latentmi.ksg.entropyd(sx, base=2)

Discrete entropy estimator sx is a list of samples

latentmi.ksg.centropyd(x, y, base=2)

The classic K-L k-nearest neighbor continuous entropy estimator for the entropy of X conditioned on Y.

latentmi.ksg.midd(x, y, base=2)

Discrete mutual information estimator Given a list of samples which can be any hashable object

latentmi.ksg.out_of_sample_avgdigamma(obs_points, est_points, dvec)

here we compute psi(nx) for query points that are not necessarily part of observed samples

Parameters:

• obs_points

• est_points

• dvec

latentmi.ksg.out_of_sample_pmi(obs_x, obs_y, est_x, est_y, k=3, base=2)