Package: RSSL 0.9.7

Jesse Krijthe

RSSL: Implementations of Semi-Supervised Learning Approaches for Classification

A collection of implementations of semi-supervised classifiers and methods to evaluate their performance. The package includes implementations of, among others, Implicitly Constrained Learning, Moment Constrained Learning, the Transductive SVM, Manifold regularization, Maximum Contrastive Pessimistic Likelihood estimation, S4VM and WellSVM.

Authors:Jesse Krijthe [aut, cre]

RSSL_0.9.7.tar.gz
RSSL_0.9.7.tar.gz(r-4.5-noble)RSSL_0.9.7.tar.gz(r-4.4-noble)
RSSL_0.9.7.tgz(r-4.4-emscripten)RSSL_0.9.7.tgz(r-4.3-emscripten)
RSSL.pdf |RSSL.html
RSSL/json (API)

# Install 'RSSL' in R:
install.packages('RSSL', repos = c('https://cran.r-universe.dev', 'https://cloud.r-project.org'))

Peer review:

Bug tracker:https://github.com/jkrijthe/rssl/issues

Uses libs:
  • openblas– Optimized BLAS
  • c++– GNU Standard C++ Library v3
Datasets:
  • diabetes - Diabetes data for unit testing
  • svmlin_example - Test data from the svmlin implementation
  • testdata - Example semi-supervised problem
  • wdbc - Wdbc data for unit testing

openblascpp

3.76 score 3 stars 1 packages 128 scripts 286 downloads 1 mentions 75 exports 43 dependencies

Last updated 1 years agofrom:f824f383a6. Checks:OK: 2. Indexed: no.

TargetResultDate
Doc / VignettesOKDec 02 2024
R-4.5-linux-x86_64OKDec 02 2024

Exports:add_missinglabels_marBaseClassifierclapplycov_mlCrossValidationSSLdecisionvaluesdf_to_matricesEMLeastSquaresClassifierEMLinearDiscriminantClassifierEMNearestMeanClassifierEntropyRegularizedLogisticRegressiongenerate2ClassGaussiangenerateABAgenerateCrescentMoongenerateFourClustersgenerateParallelPlanesgenerateSlicedCookiegenerateSpiralsgenerateTwoCirclesgeom_classifiergeom_linearclassifierGRFClassifierICLeastSquaresClassifierICLinearDiscriminantClassifierKernelICLeastSquaresClassifierKernelLeastSquaresClassifierLaplacianKernelLeastSquaresClassifierLaplacianSVMLearningCurveSSLLeastSquaresClassifierline_coefficientsLinearDiscriminantClassifierLinearSVMLinearTSVMLogisticLossClassifierLogisticRegressionLogisticRegressionFastlogsumexplosslosslogsumlosspartMajorityClassClassifierMCLinearDiscriminantClassifierMCNearestMeanClassifierMCPLDAmeasure_accuracymeasure_errormeasure_losslabmeasure_losstestmeasure_losstrainmissing_labelsNearestMeanClassifierposteriorpredictPreProcessingPreProcessingPredictQuadraticDiscriminantClassifierresponsibilitiesS4VMsample_k_per_levelscaleMatrixSelfLearningshowsolve_svmsplit_dataset_sslsplit_randomSSLDataFrameToMatricesstat_classifierstderrorSVMsvmlintrue_labelsTSVMUSMLeastSquaresClassifierWellSVM

Dependencies:cliclustercolorspacecpp11dplyrfansifarvergenericsggplot2gluegtableisobandkernlablabelinglatticelifecyclemagrittrMASSMatrixmgcvmunsellnlmepillarpkgconfigplyrpurrrquadprogR6RColorBrewerRcppRcppArmadilloreshape2rlangscalesstringistringrtibbletidyrtidyselectutf8vctrsviridisLitewithr

Readme and manuals

Help Manual

Help pageTopics
Throw out labels at randomadd_missinglabels_mar
Calculate knn adjacency matrixadjacency_knn
Classifier used for enabling shared documenting of parametersBaseClassifier
Merge result of cross-validation runs on single datasets into a the same objectc.CrossValidation
Use mclapply conditional on not being in RStudioclapply
Biased (maximum likelihood) estimate of the covariance matrixcov_ml
Cross-validation in semi-supervised settingCrossValidationSSL CrossValidationSSL.list CrossValidationSSL.matrix
Decision values returned by a classifier for a set of objectsdecisionvalues decisionvalues,KernelLeastSquaresClassifier-method decisionvalues,LeastSquaresClassifier-method decisionvalues,LinearSVM-method decisionvalues,SVM-method decisionvalues,svmlinClassifier-method decisionvalues,TSVM-method
Convert data.frame with missing labels to matricesdf_to_matrices
diabetes data for unit testingdiabetes
An Expectation Maximization like approach to Semi-Supervised Least Squares ClassificationEMLeastSquaresClassifier
Semi-Supervised Linear Discriminant Analysis using Expectation MaximizationEMLinearDiscriminantClassifier
Semi-Supervised Nearest Mean Classifier using Expectation MaximizationEMNearestMeanClassifier
Entropy Regularized Logistic RegressionEntropyRegularizedLogisticRegression
Find a violated labelfind_a_violated_label
calculated the gaussian kernel matrixgaussian_kernel
Generate data from 2 Gaussian distributed classesgenerate2ClassGaussian
Generate data from 2 alternating classesgenerateABA
Generate Crescent Moon datasetgenerateCrescentMoon
Generate Four Clusters datasetgenerateFourClusters
Generate Parallel planesgenerateParallelPlanes
Generate Sliced Cookie datasetgenerateSlicedCookie
Generate Intersecting SpiralsgenerateSpirals
Generate data from 2 circlesgenerateTwoCircles
Plot RSSL classifier boundary (deprecated)geom_classifier
Plot linear RSSL classifier boundarygeom_linearclassifier
Label propagation using Gaussian Random Fields and Harmonic functionsGRFClassifier
Direct R Translation of Xiaojin Zhu's Matlab code to determine harmonic solutionharmonic_function
Implicitly Constrained Least Squares ClassifierICLeastSquaresClassifier
Implicitly Constrained Semi-supervised Linear Discriminant ClassifierICLinearDiscriminantClassifier
Kernelized Implicitly Constrained Least Squares ClassificationKernelICLeastSquaresClassifier
Kernelized Least Squares ClassifierKernelLeastSquaresClassifier
Laplacian Regularized Least Squares ClassifierLaplacianKernelLeastSquaresClassifier
Laplacian SVM classifierLaplacianSVM
Compute Semi-Supervised Learning CurveLearningCurveSSL LearningCurveSSL.matrix
Least Squares ClassifierLeastSquaresClassifier
Loss of a classifier or regression functionline_coefficients line_coefficients,LeastSquaresClassifier-method line_coefficients,LinearSVM-method line_coefficients,LogisticLossClassifier-method line_coefficients,LogisticRegression-method line_coefficients,NormalBasedClassifier-method line_coefficients,QuadraticDiscriminantClassifier-method line_coefficients,SelfLearning-method
Linear Discriminant ClassifierLinearDiscriminantClassifier
Linear SVM ClassifierLinearSVM
LinearSVM ClassLinearSVM-class
Linear CCCP Transductive SVM classifierLinearTSVM
Local descentlocalDescent
Logistic Loss ClassifierLogisticLossClassifier
LogisticLossClassifierLogisticLossClassifier-class
(Regularized) Logistic Regression implementationLogisticRegression
Logistic Regression implementation that uses R's glmLogisticRegressionFast
Numerically more stable way to calculate log sum explogsumexp
Loss of a classifier or regression functionloss loss,KernelLeastSquaresClassifier-method loss,LeastSquaresClassifier-method loss,LinearSVM-method loss,LogisticLossClassifier-method loss,LogisticRegression-method loss,MajorityClassClassifier-method loss,NormalBasedClassifier-method loss,SelfLearning-method loss,SVM-method loss,svmlinClassifier-method loss,USMLeastSquaresClassifier-method
LogsumLoss of a classifier or regression functionlosslogsum losslogsum,NormalBasedClassifier-method
Loss of a classifier or regression function evaluated on partial labelslosspart losspart,NormalBasedClassifier-method
Majority Class ClassifierMajorityClassClassifier
Moment Constrained Semi-supervised Linear Discriminant Analysis.MCLinearDiscriminantClassifier
Moment Constrained Semi-supervised Nearest Mean ClassifierMCNearestMeanClassifier
Maximum Contrastive Pessimistic Likelihood Estimation for Linear Discriminant AnalysisMCPLDA
Performance measures used in classifier evaluationmeasure_accuracy measure_error measure_losslab measure_losstest measure_losstrain
Implements weighted likelihood estimation for LDAminimaxlda
Access the true labels for the objects with missing labels when they are stored as an attribute in a data framemissing_labels
Nearest Mean ClassifierNearestMeanClassifier
Plot CrossValidation objectplot.CrossValidation
Plot LearningCurve objectplot.LearningCurve
Class Posteriors of a classifierposterior posterior,LogisticRegression-method posterior,NormalBasedClassifier-method
Predict for matrix scaling inspired by stdize from the PLS packagepredict,scaleMatrix-method
Preprocess the input to a classification functionPreProcessing
Preprocess the input for a new set of test objects for classifierPreProcessingPredict
Print CrossValidation objectprint.CrossValidation
Print LearningCurve objectprint.LearningCurve
Project an n-dim vector y to the simplex Dnprojection_simplex
Quadratic Discriminant ClassifierQuadraticDiscriminantClassifier
Responsibilities assigned to the unlabeled objectsresponsibilities
Show RSSL classifierrssl-formatting show,Classifier-method show,NormalBasedClassifier-method show,scaleMatrix-method
Predict using RSSL classifierdecisionvalues,WellSVM-method predict,GRFClassifier-method predict,KernelLeastSquaresClassifier-method predict,LeastSquaresClassifier-method predict,LinearSVM-method predict,LogisticLossClassifier-method predict,LogisticRegression-method predict,MajorityClassClassifier-method predict,NormalBasedClassifier-method predict,SelfLearning-method predict,SVM-method predict,svmlinClassifier-method predict,USMLeastSquaresClassifier-method predict,WellSVM-method responsibilities,GRFClassifier-method rssl-predict
Safe Semi-supervised Support Vector Machine (S4VM)S4VM
LinearSVM ClassS4VM-class
Sample k indices per levels from a factorsample_k_per_level
Matrix centering and scalingscaleMatrix
Self-Learning approach to Semi-supervised LearningSelfLearning
SVM solve.QP implementationsolve_svm
Create Train, Test and Unlabeled Setsplit_dataset_ssl
Randomly split dataset in multiple partssplit_random
Convert data.frame to matrices for semi-supervised learnersSSLDataFrameToMatrices
Plot RSSL classifier boundariesstat_classifier
Calculate the standard error of the mean from a vector of numbersstderror
Summary of Crossvalidation resultssummary.CrossValidation
Inverse of a matrix using the singular value decompositionsvdinv
Taking the inverse of the square root of the matrix using the singular value decompositionsvdinvsqrtm
Taking the square root of a matrix using the singular value decompositionsvdsqrtm
SVM ClassifierSVM
svmlin implementation by Sindhwani & Keerthi (2006)svmlin
Test data from the svmlin implementationsvmlin_example
Train SVMsvmproblem
Example semi-supervised problemtestdata
Refine the prediction to satisfy the balance constraintthreshold
Access the true labels when they are stored as an attribute in a data frametrue_labels
Transductive SVM classifier using the convex concave procedureTSVM
Updated Second Moment Least Squares ClassifierUSMLeastSquaresClassifier
USMLeastSquaresClassifierUSMLeastSquaresClassifier-class
wdbc data for unit testingwdbc
WellSVM for Semi-supervised LearningWellSVM
wellsvm implements the wellsvm algorithm as shown in [1].wellsvm_direct
Convex relaxation of S3VM by label generationWellSVM_SSL
A degenerated version of WellSVM where the labels are complete, that is, supervised learningWellSVM_supervised
Implements weighted likelihood estimation for LDAwlda
Measures the expected error of the LDA model defined by m, p, and iW on the data set a, where weights w are potentially taken into accountwlda_error
Measures the expected log-likelihood of the LDA model defined by m, p, and iW on the data set a, where weights w are potentially taken into accountwlda_loglik