API Documentation

Main classes that implement the post-hoc adaptation framework

Workbench(model[, cov, pattern_modifier, …])

Work bench for post-hoc alteration of a linear model.

WorkbenchOptimizer(model[, cov, …])

Work bench for post-hoc alteration of a linear model, with optimization.

Beamformer(template[, center, normalize, …])

A beamformer filter.

Ways of estimating the covariance matrix

Empirical()

Empirical estimation of the covariance matrix.

L2([alpha, scale_by_var])

Estimation of the covariance matrix using L2 regularization.

L2Kernel([alpha, scale_by_var])

L2 estimation of the covariance using the kernel formulation.

Shrinkage([alpha])

Estimation of the covariance matrix using Shrinkage regularization.

ShrinkageKernel([alpha])

Shrinkage estimation of the covariance using the kernel formulation.

KroneckerKernel([outer_size, inner_size, …])

Kronecker shrinkage of the covariance using the kernel formulation.

Utilities for efficient leave-one-out crossvalidation

loo(X[, axis])

Generates leave-one-out selections along a given axis.

loo_kern_inv(K)

A generator for leave-one-out crossval iterations from a kernel matrix.

loo_mean_norm(X[, return_norm])

Compute the mean() and L2 norm of a matrix in a leave-one-out manner.

loo_ols_regression(X, y[, normalize, …])

Generate OLS regression coefficients for leave-one-out iterations.

loo_ols_values(X, y[, normalize, fit_intercept])

Generate OLS regression values for leave-one-out iterations.

loo_patterns_from_model(model, X, y[, …])

Generate patterns for leave-one-out iterations of the given model.

Scoring functions

correlation_score(model, X, y)

Pearson correlation score.

logistic_loss_score(model, X, y)

Logistic loss score.