This is a utility class to use online training algorithms from a data set transparently as a batch training algorithm.
| data_tmpl | The type of a data set. | |
| trainer_tmpl | The type of an online training algorithm. |
Public Types | |
| typedef data_tmpl | data_type |
| The type representing a data set for training. | |
| typedef trainer_tmpl | trainer_type |
| The type implementing a training algorithm. | |
| typedef data_type::const_iterator | const_iterator |
| A type providing a read-only random-access iterator for instances. | |
| typedef data_type::instance_type | instance_type |
| The type representing an instance in the training data. | |
| typedef instance_type::attribute_type | attribute_type |
| The type representing an attribute in an instance. | |
| typedef instance_type::value_type | value_type |
| The type representing a value. | |
| typedef trainer_type::error_type | error_type |
| The type implementing an error function. | |
| typedef trainer_type::model_type | model_type |
| The type implementing a model (weight vector for features). | |
Public Member Functions | |
| online_scheduler_binary () | |
| Constructs the object. | |
| virtual | ~online_scheduler_binary () |
| Destructs the object. | |
| void | clear () |
| Resets the internal states and parameters to default. | |
| parameter_exchange & | params () |
| Obtains the parameter interface. | |
| const model_type & | model () const |
| Obtains a read-only access to the weight vector (model). | |
| void | train (const data_type &data, std::ostream &os, int holdout=-1, bool acconly=true) |
| Trains a model on a data set. | |
Protected Attributes | |
| trainer_type | m_trainer |
| Trainer type. | |
| std::string | m_sample |
| The sample method. | |
| int | m_max_iterations |
| The maximum number of iterations. | |
| value_type | m_c |
| The parameter for regularization. | |
| int | m_period |
| The period to measure the improvement ratio of loss. | |
| value_type | m_epsilon |
| The epsilon for improvement ratio. | |
| parameter_exchange& classias::train::online_scheduler_binary< data_tmpl, trainer_tmpl >::params | ( | ) | [inline] |
Obtains the parameter interface.
| const model_type& classias::train::online_scheduler_binary< data_tmpl, trainer_tmpl >::model | ( | ) | const [inline] |
Obtains a read-only access to the weight vector (model).
| void classias::train::online_scheduler_binary< data_tmpl, trainer_tmpl >::train | ( | const data_type & | data, | |
| std::ostream & | os, | |||
| int | holdout = -1, |
|||
| bool | acconly = true | |||
| ) | [inline] |
Trains a model on a data set.
| data | The data set for training (and holdout evaluation). | |
| os | The output stream for progress reports. | |
| holdout | The group number for holdout evaluation. Specify a negative value if a holdout evaluation is unnecessary. | |
| acconly | Unused (reserved only for the compatibility with multi-class classification). |