{"Package":"RSSL","Version":"0.9.3","Title":"Implementations of Semi-Supervised Learning Approaches for\nClassification","Authors@R":"c(person(\"Jesse\", \"Krijthe\", role = c(\"aut\", \"cre\"),\nemail = \"jkrijthe@gmail.com\"))","Depends":"R(>= 2.10.0)","Imports":"methods, Rcpp, MASS, kernlab, quadprog, Matrix, dplyr, tidyr,\nggplot2, reshape2, scales, cluster","LinkingTo":"Rcpp, RcppArmadillo","Suggests":"testthat, rmarkdown, SparseM, numDeriv, LiblineaR","Description":"A collection of implementations of semi-supervised classifiers\nand methods to evaluate their performance. The package includes implementations\nof, among others, Implicitly Constrained Learning, Moment Constrained Learning,\nthe Transductive SVM, Manifold regularization, Maximum Contrastive Pessimistic\nLikelihood estimation, S4VM and WellSVM.","License":"GPL (>= 2)","URL":"https://github.com/jkrijthe/RSSL","BugReports":"https://github.com/jkrijthe/RSSL","Collate":"'Generics.R' 'Classifier.R' 'CrossValidation.R'\n'LeastSquaresClassifier.R' 'EMLeastSquaresClassifier.R'\n'NormalBasedClassifier.R' 'LinearDiscriminantClassifier.R'\n'EMLinearDiscriminantClassifier.R' 'NearestMeanClassifier.R'\n'EMNearestMeanClassifier.R' 'LogisticRegression.R'\n'EntropyRegularizedLogisticRegression.R' 'Evaluate.R'\n'GRFClassifier.R' 'GenerateSSLData.R' 'HelperFunctions.R'\n'ICLeastSquaresClassifier.R' 'ICLinearDiscriminantClassifier.R'\n'KernelLeastSquaresClassifier.R'\n'KernelICLeastSquaresClassifier.R'\n'LaplacianKernelLeastSquaresClassifier.R' 'LaplacianSVM.R'\n'LearningCurve.R' 'LinearSVM.R' 'LogisticLossClassifier.R'\n'MCLinearDiscriminantClassifier.R' 'MCNearestMeanClassifier.R'\n'MCPLDA.R' 'MajorityClassClassifier.R' 'Measures.R'\n'Plotting.R' 'QuadraticDiscriminantClassifier.R' 'RSSL.R'\n'RcppExports.R' 'S4VM.R' 'SVM.R' 'SelfLearning.R' 'TSVM.R'\n'USMLeastSquaresClassifier.R' 'WellSVM.R' 'scaleMatrix.R'\n'svmd.R' 'svmlin.R' 'testdata-data.R'","Encoding":"UTF-8","RoxygenNote":"7.1.0","NeedsCompilation":"yes","Packaged":"2020-11-13 17:24:31 UTC; jkrijthe","Author":"Jesse Krijthe [aut, cre]","Maintainer":"Jesse Krijthe ","Repository":"CRAN","Date/Publication":"2020-11-13 18:00:10 UTC","repoType":"cran","tarballUrl":"ftp://cran.r-project.org/pub/R/src/contrib/RSSL_0.9.3.tar.gz","jsonAuthors":[{"name":"Jesse Krijthe","email":"jkrijthe@gmail.com","maintainer":true}],"readme":"\n[![Travis-CI Build\nStatus](https://travis-ci.org/jkrijthe/RSSL.png?branch=master)](https://travis-ci.org/jkrijthe/RSSL)\n[![codecov.io](https://codecov.io/github/jkrijthe/RSSL/coverage.svg?branch=master)](https://codecov.io/github/jkrijthe/RSSL?branch=master)\n[![CRAN mirror\ndownloads](http://cranlogs.r-pkg.org/badges/RSSL)](https://cran.r-project.org/package=RSSL)\n\nR Semi-Supervised Learning package\n==================================\n\nThis R package provides implementations of several semi-supervised\nlearning methods, in particular, our own work involving constraint based\nsemi-supervised learning.\n\nTo cite the package, use either of these two references:\n\n- Krijthe, J. H. (2016). RSSL: R package for Semi-supervised Learning.\n In B. Kerautret, M. Colom, & P. Monasse (Eds.), Reproducible\n Research in Pattern Recognition. RRPR 2016. Lecture Notes in\n Computer Science, vol 10214. (pp. 104–115). Springer International\n Publishing.\n https://doi.org/10.1007/978-3-319-56414-2_8.\n arxiv:\n https://arxiv.org/abs/1612.07993\n- Krijthe, J.H. & Loog, M. (2015). Implicitly Constrained\n Semi-Supervised Least Squares Classification. In E. Fromont, T. de\n Bie, & M. van Leeuwen, eds. 14th International Symposium on Advances\n in Intelligent Data Analysis XIV (Lecture Notes in Computer Science\n Volume 9385). Saint Etienne. France, pp. 158-169.\n\nInstallation Instructions\n=========================\n\nThis package available on CRAN. The easiest way to install the package\nis to use:\n\n``` r\ninstall.packages(\"RSSL\")\n```\n\nTo install the latest version of the package using the devtools package:\n\n``` r\nlibrary(devtools)\ninstall_github(\"jkrijthe/RSSL\")\n```\n\nUsage\n=====\n\nAfter installation, load the package as usual:\n\n``` r\nlibrary(RSSL)\n```\n\nThe following code generates a simple dataset, trains a supervised and\ntwo semi-supervised classifiers and evaluates their performance:\n\n``` r\nlibrary(dplyr,warn.conflicts = FALSE)\nlibrary(ggplot2,warn.conflicts = FALSE)\n\nset.seed(2)\ndf <- generate2ClassGaussian(200, d=2, var = 0.2, expected=TRUE)\n\n# Randomly remove labels\ndf <- df %>% add_missinglabels_mar(Class~.,prob=0.98) \n\n# Train classifier\ng_nm <- NearestMeanClassifier(Class~.,df,prior=matrix(0.5,2))\ng_self <- SelfLearning(Class~.,df,\n method=NearestMeanClassifier,\n prior=matrix(0.5,2))\n\n# Plot dataset\ndf %>% \n ggplot(aes(x=X1,y=X2,color=Class,size=Class)) +\n geom_point() +\n coord_equal() +\n scale_size_manual(values=c(\"-1\"=3,\"1\"=3), na.value=1) +\n geom_linearclassifier(\"Supervised\"=g_nm,\n \"Semi-supervised\"=g_self)\n```\n\n![](tools/example-1.png)\n\n``` r\n\n# Evaluate performance: Squared Loss & Error Rate\nmean(loss(g_nm,df))\nmean(loss(g_self,df))\n\n\nmean(predict(g_nm,df)!=df$Class)\nmean(predict(g_self,df)!=df$Class)\n```\n\nAcknowledgement\n===============\n\nWork on this package was supported by Project 23 of the Dutch national\nprogram COMMIT.","jobInfo":{"package":"RSSL","version":"0.9.3","parsingStatus":"success","parserVersion":1,"parsedAt":"2022-01-17T16:29:48+0000"}}

Description

Usage

Arguments

Value