Kernel Methods<\/span><\/strong><\/h1>\n
\nBinary C SVM<\/h3>\n
Java class<\/strong>:\u00a0BinaryCSvmClassification<\/a><\/p>\nSource code<\/strong>:\u00a0BinaryCSvmClassification.java<\/a><\/p>\nMaven Project<\/strong>:\u00a0kelp-core<\/p>\nJSON type<\/strong>:\u00a0binaryCSvmClassification<\/p>\nDescription<\/strong>:\u00a0It is the KeLP implementation of C-Support Vector Machine learning algorithm [Cortes and Vapnik(1995)]. It is a learning algorithm for binary classification and it relies on kernel functions. It is a Java porting of the library LIBSVM v3.17, written in C++ [Chang and Lin(2011)]<\/p>\nParameters:<\/strong><\/p>\n\n- kernel<\/em>: The kernel function<\/li>\n
- label<\/em>: The label to be learned<\/li>\n
- cp<\/em>: The regularization parameter for positive examples<\/li>\n
- cn<\/em>: The regularization parameter for negative examples<\/li>\n
- useFairness<\/em>: A boolean parameter to force the fairness policy<\/li>\n<\/ul>\n
\nBinary \u03bd-SVM<\/h3>\n
Java class<\/strong>:\u00a0BinaryNuSvmClassification<\/a><\/p>\nSource code<\/strong>:\u00a0BinaryNuSvmClassification.java<\/a><\/p>\nMaven Project<\/strong>:\u00a0kelp-core<\/p>\nJSON type<\/strong>:\u00a0binaryNuSvmClassification<\/p>\nDescription<\/strong>:\u00a0It is the KeLP implementation of the \u03bd-Support Vector Machine learning algorithm [Scholkopf et al.(2000)]. It is a learning algorithm for binary classification and it relies on kernel functions. It is a Java porting of the library LIBSVM v3.17, written in C++ [Chang and Lin(2011)].<\/p>\nParameters:<\/strong><\/p>\n\n- kernel<\/em>: The kernel function<\/li>\n
- label<\/em>: The label to be learned<\/li>\n
- nu<\/em>: The \u03bd<\/em> parameter, i.e., the percentage of training example to be used as Support Vectors<\/li>\n
- useFairness<\/em>: A boolean parameter to force the fairness policy<\/li>\n<\/ul>\n
\nOne-Class SVM<\/h3>\n
Java class<\/strong>:\u00a0OneClassSvmClassification<\/a><\/p>\nSource code<\/strong>:\u00a0OneClassSvmClassification.java<\/a><\/p>\nMaven Project<\/strong>: kelp-core<\/p>\nJSON type<\/strong>:\u00a0oneClassSvmClassification<\/p>\nDescription<\/strong>:\u00a0It is KeLP implementation of One-Class Support Vector Machine learning algorithm [Scholkopf et al.(2001)]. It is a learning algorithm for estimating the Support of a HighDimensional Distribution and it relies on kernel functions. The model is acquired only by considering positive examples. It is useful in anomaly detection (a.k.a. novelty detection). It is a Java porting of the library LIBSVM v3.17, written in C++ [Chang and Lin(2011)]<\/p>\nParameters:<\/strong><\/p>\n\n- kernel<\/em>: The kernel function<\/li>\n
- label<\/em>: The label to be learned<\/li>\n
- nu<\/em>: The \u03bd parameter, i.e., the percentage of training example to be used as Support Vectors<\/li>\n<\/ul>\n
\nKernel-based Perceptron<\/h3>\n
Java class<\/strong>:\u00a0KernelizedPerceptron<\/a><\/p>\nSource code<\/strong>:\u00a0KernelizedPerceptron.java<\/a><\/p>\nMaven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\nJSON type<\/strong>:\u00a0kernelizedPerceptron<\/p>\nDescription<\/strong>:\u00a0The perceptron learning algorithm for classification tasks (Kernel machine version, presented in [Rosenblat(1958)]).<\/p>\nParameters:<\/strong><\/p>\n\n- kernel<\/em>: The kernel function<\/li>\n
- label<\/em>: The label to be learned<\/li>\n
- alpha<\/em>: The learning rate, i.e., the weight associated to misclassified examples during the learning process<\/li>\n
- margin<\/em>: The minimum distance from the hyperplane that an example must have in order to be not considered misclassified<\/li>\n
- unbiased<\/em>: This boolean parameter determines the use of bias b in the classification function
![\"f(x)=wx + b\"](\"http:\/\/www.kelp-ml.org\/wp-content\/ql-cache\/quicklatex.com-e3e97d9ecf31e6006910b4b8ec1c509d_l3.png\" "\\"Rendered")
. If usebias is set to true the bias is set to 0.<\/li>\n<\/ul>\n
\nKernel-based Passive Aggressive (PA) Classification<\/h3>\n
Java class<\/strong>:\u00a0KernelizedPassiveAggressiveClassification<\/a><\/p>\nSource code<\/strong>:\u00a0KernelizedPassiveAggressiveClassification.java<\/a><\/p>\nMaven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\nJSON type<\/strong>:\u00a0kernelizedPA<\/p>\nDescription<\/strong>:\u00a0Online Passive-Aggressive Learning Algorithm for classification tasks with Kernels (presented in [Crammer et al.(2006)] and extended in [Filice et al.(2014)])). Every time an example is misclassified it is added as support vector, with the weight that solves the passive aggressive minimization problem.<\/p>\nParameters:<\/strong><\/p>\n\n- label<\/em>: The label to be learned<\/li>\n
- kernel<\/em>: The kernel function<\/li>\n
- loss<\/em>: The loss function to weight each misclassification<\/li>\n
- policy<\/em>: The updating policy applied by the Passive Aggressive Algorithm when a miss-prediction occurs<\/li>\n
- cp<\/em>: The aggressiveness parameter for positive examples<\/li>\n
- cn<\/em>: The aggressiveness parameter for negative examples<\/li>\n
- useFairness<\/em>: A boolean parameter to force the fairness policy<\/li>\n<\/ul>\n
\nBudgeted Online Learning: Budgeted Passive Aggressive Classification<\/h3>\n
Java class<\/strong>:\u00a0BudgetedPassiveAggressiveClassification<\/a><\/p>\nSource code<\/strong>:\u00a0BudgetedPassiveAggressiveClassification.java<\/a><\/p>\nMaven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\nJSON type<\/strong>:\u00a0budgetedPA<\/p>\nDescription<\/strong>:\u00a0It is the implementation of the Budgeted Passive Aggressive Algorithm proposed in [Wang and Vucetic(2010)]. When the budget is full, the schema proposed in [Wang and Vucetic(2010)] to update examples (and weights) is adopted.<\/p>\nParameters:<\/strong><\/p>\n\n- label<\/em>: The label to be learned<\/li>\n
- label<\/em>: The label to be learned<\/li>\n
- kernel<\/em>: The kernel function<\/li>\n
- loss<\/em>: The loss function to weight each misclassification<\/li>\n
- policy<\/em>: The updating policy applied by the Passive Aggressive Algorithm when a miss-prediction occurs<\/li>\n
- deletingPolicy<\/em>: The policy for the removal of examples from the budget before adding new examples. This can be\n
\n- BPA S<\/em>: Budgeted Passive Aggressive Simple: when a new support vector must be added, one is removed and the weight of the other support vectors is kept unchanged<\/li>\n
- BPA 1NN<\/em>: Budgeted Passive Aggressive Nearest Neighbor: when a new support vector must be added, one is removed and the weight of its nearest neighbor is adjusted<\/li>\n<\/ul>\n<\/li>\n
- cp<\/em>: The aggressiveness parameter for positive examples<\/li>\n
- cn<\/em>: The aggressiveness parameter for negative examples<\/li>\n
- useFairness<\/em>: A boolean parameter to force the fairness policy<\/li>\n
- budget<\/em>: The maximum number of support vectors allowed in the budget<\/li>\n<\/ul>\n
\n Linear Methods<\/span><\/strong><\/h1>\n
\nLibLinear SVM Classification<\/h3>\n
Java class<\/strong>:\u00a0LibLinearLearningAlgorithm<\/a><\/p>\nSource code<\/strong>:\u00a0LibLinearLearningAlgorithm.java<\/a><\/p>\nMaven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\nJSON type<\/strong>:\u00a0liblinear<\/p>\nDescription<\/strong>:\u00a0This class implements linear SVMs models trained using a coordinate descent algorithm [Fan et al.(2008)]. It operates in an explicit feature space (i.e., it does not rely on any kernel). This code has been adapted from the Java port of the original LIBLINEAR C++ sources.<\/p>\nParameters:<\/strong><\/p>\n\n- label<\/em>: The label to be learned<\/li>\n
- cp<\/em>: The regularization parameter for positive examples<\/li>\n
- cn<\/em>: The regularization parameter for negative examples<\/li>\n
- representation<\/em>: The identifier of the representation to be considered for the training step<\/li>\n<\/ul>\n
\nPegasos Classification<\/h3>\n
Java class<\/strong>:\u00a0PegasosLearningAlgorithm<\/a><\/p>\nSource code<\/strong>:\u00a0PegasosLearningAlgorithm.java<\/a><\/p>\nMaven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\nJSON type<\/strong>:\u00a0pegasos<\/p>\nDescription<\/strong>:\u00a0It implements the Primal Estimated sub-GrAdient SOlver (PEGASOS) for SVM. It is a learning algorithm for binary linear classification Support Vector Machines. It operates in an explicit feature space (i.e., it does not rely on any kernel). Further details can be found in [Shalev-Shwartz et al.(2007)].<\/p>\nParameters:<\/strong><\/p>\n\n- label<\/em>: The label to be learned<\/li>\n
- lambda<\/em>: The regularization coefficient<\/li>\n
- iterations<\/em>: The number of iterations required from the learning algorithm<\/li>\n
- k<\/em>: The number of examples k that PEGASOS exploits in its mini-batch learning approach<\/li>\n
- representation<\/em>: The identifier of the representation to be considered for the training step<\/li>\n<\/ul>\n
\nA Dual Coordinate Descent Learning Algorithm<\/h3>\n
Java class<\/strong>:\u00a0DCDLearningAlgorithm<\/a><\/p>\n
Source code<\/strong>:\u00a0BinaryCSvmClassification.java<\/a><\/p>\n Maven Project<\/strong>:\u00a0kelp-core<\/p>\n JSON type<\/strong>:\u00a0binaryCSvmClassification<\/p>\n Description<\/strong>:\u00a0It is the KeLP implementation of C-Support Vector Machine learning algorithm [Cortes and Vapnik(1995)]. It is a learning algorithm for binary classification and it relies on kernel functions. It is a Java porting of the library LIBSVM v3.17, written in C++ [Chang and Lin(2011)]<\/p>\n Parameters:<\/strong><\/p>\n Java class<\/strong>:\u00a0BinaryNuSvmClassification<\/a><\/p>\n Source code<\/strong>:\u00a0BinaryNuSvmClassification.java<\/a><\/p>\n Maven Project<\/strong>:\u00a0kelp-core<\/p>\n JSON type<\/strong>:\u00a0binaryNuSvmClassification<\/p>\n Description<\/strong>:\u00a0It is the KeLP implementation of the \u03bd-Support Vector Machine learning algorithm [Scholkopf et al.(2000)]. It is a learning algorithm for binary classification and it relies on kernel functions. It is a Java porting of the library LIBSVM v3.17, written in C++ [Chang and Lin(2011)].<\/p>\n Parameters:<\/strong><\/p>\n Java class<\/strong>:\u00a0OneClassSvmClassification<\/a><\/p>\n Source code<\/strong>:\u00a0OneClassSvmClassification.java<\/a><\/p>\n Maven Project<\/strong>: kelp-core<\/p>\n JSON type<\/strong>:\u00a0oneClassSvmClassification<\/p>\n Description<\/strong>:\u00a0It is KeLP implementation of One-Class Support Vector Machine learning algorithm [Scholkopf et al.(2001)]. It is a learning algorithm for estimating the Support of a HighDimensional Distribution and it relies on kernel functions. The model is acquired only by considering positive examples. It is useful in anomaly detection (a.k.a. novelty detection). It is a Java porting of the library LIBSVM v3.17, written in C++ [Chang and Lin(2011)]<\/p>\n Parameters:<\/strong><\/p>\n Java class<\/strong>:\u00a0KernelizedPerceptron<\/a><\/p>\n Source code<\/strong>:\u00a0KernelizedPerceptron.java<\/a><\/p>\n Maven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\n JSON type<\/strong>:\u00a0kernelizedPerceptron<\/p>\n Description<\/strong>:\u00a0The perceptron learning algorithm for classification tasks (Kernel machine version, presented in [Rosenblat(1958)]).<\/p>\n Parameters:<\/strong><\/p>\n Java class<\/strong>:\u00a0KernelizedPassiveAggressiveClassification<\/a><\/p>\n Source code<\/strong>:\u00a0KernelizedPassiveAggressiveClassification.java<\/a><\/p>\n Maven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\n JSON type<\/strong>:\u00a0kernelizedPA<\/p>\n Description<\/strong>:\u00a0Online Passive-Aggressive Learning Algorithm for classification tasks with Kernels (presented in [Crammer et al.(2006)] and extended in [Filice et al.(2014)])). Every time an example is misclassified it is added as support vector, with the weight that solves the passive aggressive minimization problem.<\/p>\n Parameters:<\/strong><\/p>\n Java class<\/strong>:\u00a0BudgetedPassiveAggressiveClassification<\/a><\/p>\n Source code<\/strong>:\u00a0BudgetedPassiveAggressiveClassification.java<\/a><\/p>\n Maven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\n JSON type<\/strong>:\u00a0budgetedPA<\/p>\n Description<\/strong>:\u00a0It is the implementation of the Budgeted Passive Aggressive Algorithm proposed in [Wang and Vucetic(2010)]. When the budget is full, the schema proposed in [Wang and Vucetic(2010)] to update examples (and weights) is adopted.<\/p>\n Parameters:<\/strong><\/p>\n Java class<\/strong>:\u00a0LibLinearLearningAlgorithm<\/a><\/p>\n Source code<\/strong>:\u00a0LibLinearLearningAlgorithm.java<\/a><\/p>\n Maven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\n JSON type<\/strong>:\u00a0liblinear<\/p>\n Description<\/strong>:\u00a0This class implements linear SVMs models trained using a coordinate descent algorithm [Fan et al.(2008)]. It operates in an explicit feature space (i.e., it does not rely on any kernel). This code has been adapted from the Java port of the original LIBLINEAR C++ sources.<\/p>\n Parameters:<\/strong><\/p>\n Java class<\/strong>:\u00a0PegasosLearningAlgorithm<\/a><\/p>\n Source code<\/strong>:\u00a0PegasosLearningAlgorithm.java<\/a><\/p>\n Maven Project<\/strong>:\u00a0kelp-additional-algorithms<\/p>\n JSON type<\/strong>:\u00a0pegasos<\/p>\n Description<\/strong>:\u00a0It implements the Primal Estimated sub-GrAdient SOlver (PEGASOS) for SVM. It is a learning algorithm for binary linear classification Support Vector Machines. It operates in an explicit feature space (i.e., it does not rely on any kernel). Further details can be found in [Shalev-Shwartz et al.(2007)].<\/p>\n Parameters:<\/strong><\/p>\n Java class<\/strong>:\u00a0DCDLearningAlgorithm<\/a><\/p>\n\n
\nBinary \u03bd-SVM<\/h3>\n
\n
\nOne-Class SVM<\/h3>\n
\n
\nKernel-based Perceptron<\/h3>\n
\n
. If usebias is set to true the bias is set to 0.<\/li>\n<\/ul>\n
\nKernel-based Passive Aggressive (PA) Classification<\/h3>\n
\n
\nBudgeted Online Learning: Budgeted Passive Aggressive Classification<\/h3>\n
\n
\n
\n Linear Methods<\/span><\/strong><\/h1>\n
\nLibLinear SVM Classification<\/h3>\n
\n
\nPegasos Classification<\/h3>\n
\n
\nA Dual Coordinate Descent Learning Algorithm<\/h3>\n
![\"L $^1$\"](\"http:\/\/www.kelp-ml.org\/wp-content\/ql-cache\/quicklatex.com-4bcbda644029cb93675c1aef8486449f_l3.png\" "\\"Rendered")
or![\"L$^ 2$\"](\"http:\/\/www.kelp-ml.org\/wp-content\/ql-cache\/quicklatex.com-9f84f841a6a7b22155628eb6102a9774_l3.png\" "\\"Rendered")
Support Vector Machine for binary classification [Hsieh et al.(2008)].<\/p>\n![\"b\"](\"http:\/\/www.kelp-ml.org\/wp-content\/ql-cache\/quicklatex.com-f56d50c26583f9a035ff6b4e3c0ca5c0_l3.png\" "\\"Rendered")
in the classification function ![\"f(x)=w(x)+b\"](\"http:\/\/www.kelp-ml.org\/wp-content\/ql-cache\/quicklatex.com-a0991460c816e2f2285b43f30842f344_l3.png\" "\\"Rendered")
. If usebias is set to false the bias is set to 0<\/li>\n