Package 'MAKL'

Title: Multiple Approximate Kernel Learning (MAKL)
Description: R package associated with the Multiple Approximate Kernel Learning (MAKL) algorithm proposed in <doi:10.1093/bioinformatics/btac241>. The algorithm fits multiple approximate kernel learning (MAKL) models that are fast, scalable and interpretable.
Authors: Ayyüce Begüm Bektaş [aut, cre] , Mehmet Gönen [aut]
Maintainer: Ayyüce Begüm Bektaş <[email protected]>
License: GPL (>= 3)
Version: 1.0.1
Built: 2025-02-22 03:23:51 UTC
Source: https://github.com/cran/MAKL

Help Index

Test the Multiple Approximate Kernel Learning (MAKL) Model

Description

Binary classification of the test data, using the MAKL model resulted from makl_train().

Usage

makl_test(X, y, makl_model)

Arguments

X

test dataset, matrix of size T x d.
y

response vector of length T, containing only -1 and 1.
makl_model

a list containing the MAKL model returning from makl_train().

Value

a list containing the predictions for test instances and the area under the ROC curve (AUROC) values with corresponding number of used kernels for prediction.

Train a Multiple Approximate Kernel Learning (MAKL) Model

Description

Train a MAKL model to be used as an input to makl_test().

Usage

makl_train(
  X,
  y,
  D = 100,
  sigma_N = 1000,
  CV = 1,
  lambda_set = c(0.9, 0.8, 0.7, 0.6),
  membership
)

Arguments

X

training dataset, matrix of size N x d.
y

response vector of length N, containing only -1 and 1.
D

numeric value related to the number of random features to be used for approximation.
sigma_N

numeric value preferably smaller than N, used to calculate sigma to create random features.
CV

integer value between 0 and N. If CV is equal to 0 or 1, no cross validation is performed. If CV is greater than or equal to 2, CV is assigned as fold count in the cross validation.
lambda_set

a continuous number between 0 and 1, used for regularization.
membership

a list of length of number of groups, containing feature memberships to each group.

Value

a list containing the MAKL model and related parameters to be used in makl_test().