Classias: C++ Template Class Library for Classification

Sample programs

Class reference

Instance types
- Binary instance: classias::binary_instance_base
- Multi-class instance: classias::multi_instance_base
- Candidate instance: classias::candidate_instance_base
Data set
- Binary data set: classias::binary_data_base
- Binary data set with a string quark for attributes: classias::binary_data_with_quark_base
- Multi-class data set: classias::multi_data_base
- Multi-class data set with string quarks for attributes and labels: classias::multi_data_with_quark_base
- Candidate data set: classias::candidate_data_base
- Candidate data set with string quarks for attributes and labels: classias::candidate_data_with_quark_base
Feature generators
- Dummy feature generator for candidate instances: classias::thru_feature_generator_base
- Dense feature generator: classias::dense_feature_generator_base
- Sparse feature generator: classias::sparse_feature_generator_base
Classifiers and error functions
- Linear binary classifier: classias::classify::linear_binary
- Linear binary classifier with logistic-sigmoid loss function: classias::classify::linear_binary_logistic
- Linear binary classifier with hinge loss function: classias::classify::linear_binary_hinge
- Linear multi classifier: classias::classify::linear_multi
- Linear multi classifier with soft-max function: classias::classify::linear_binary_logistic
Batch training algorithms
- Gradient descent with L-BFGS/OW-LQN for logistic regression: classias::train::lbfgs_logistic_binary
- Gradient descent with L-BFGS/OW-LQN for multi-class logistic regression: classias::train::lbfgs_logistic_multi
- Scheduler for applying an online-training algorithm to a data set for binary classification: classias::train::online_scheduler_binary
- Scheduler for applying an online-training algorithm to a data set for multi classification: classias::train::online_scheduler_multi
Online training algorithms
- Averaged perceptron for binary classification: classias::train::averaged_perceptron_binary
- Averaged perceptron for multi/candidate classification: classias::train::averaged_perceptron_multi
- Pegasos for binary classification: classias::train::pegasos_binary
- Pegasos for multi/candidate classification: classias::train::pegasos_multi
- Truncated gradient for binary classification: classias::train::truncated_gradient_binary
- Truncated gradient for multi/candidate classification: classias::train::truncated_gradient_multi
Basic data types
- Instance weight: classias::weight_base
- Instance group number: classias::group_base
- Sparse vector: classias::sparse_vector_base
- Quark with one item (item-to-integer mapping): classias::quark_base
- Quark with two items (item-pair-to-integer mapping): classias::quark2_base
- Quark exception: classias::quark_error
Miscellaneous utilities
- Accuracy counter: classias::accuracy
- Performance evaluation with precision, recall, and F1 scores classias::precall
- Parameter exchange classias::parameter_exchange
- Exception class for parameter exchange classias::unknown_parameter