<< Click to Display Table of Contents >> Navigation:  Reference Manual > Learning > DSL_bsEvaluator

Header file: bs.h

---

virtual int Evaluate(int iteration, double bsScore, double bestScore,

 DSL_network &net, const DSL_dataset &ds,

 const std::vector<DSL_datasetMatch> &matching,

 DSL_progress *progress,

 double &outputScore) = 0;

DSL_bsEvaluator class declares a single pure virtual function that must be overridden in the derived classes. The Evaluate method should return an outputScore for the net using the data set ds (the learning data set passed as input to DSL_bs::Learn). A possible implementation of Evaluate can calculate the prediction accuracy using K-fold cross-validation and return the calculated accuracy as the output score. However, any algorithm can be used with any additional data available. The additional data should be stored as members of the class derived from DSL_bsEvaluator.