Vita
|
Dynamic training Subset Selection. More...
#include <dss.h>
Inherits vita::validation_strategy.
Public Member Functions | |
void | close (unsigned) override |
Moves all the example in the validation set. More... | |
dss (src_problem &, cached_evaluator &, cached_evaluator &) | |
Sets up a DSS validator. More... | |
void | init (unsigned) override |
Available examples are randomly partitioned into two independent sets according to a given percentage. More... | |
bool | shake (unsigned) override |
Changes the training environment. More... | |
virtual void | close (unsigned) |
De-initializes the data structures needed for the validation strategy. More... | |
virtual void | init (unsigned)=0 |
Initializes the data structures needed for the validation strategy. More... | |
virtual bool | shake (unsigned) |
Changes the training environment. More... | |
Dynamic training Subset Selection.
When using GP on a difficult problem, with a large set of training data, a large population size is needed and a very large number of evaluations must be carried out. DSS is a subset selection method which uses the available information to select:
vita::dss::dss | ( | src_problem & | prob, |
cached_evaluator & | eva_t, | ||
cached_evaluator & | eva_v | ||
) |
Sets up a DSS validator.
[in] | prob | current problem |
[in] | eva_t | active training evaluator |
[in] | eva_v | active validation evaluator |
|
overridevirtual |
Moves all the example in the validation set.
Reimplemented from vita::validation_strategy.
|
overridevirtual |
Available examples are randomly partitioned into two independent sets according to a given percentage.
Implements vita::validation_strategy.
|
overridevirtual |
Changes the training environment.
true
if the training environment has changedBy default does nothing, signalling that nothing is changed.
Reimplemented from vita::validation_strategy.