C++ API Reference for Intel® Data Analytics Acceleration Library 2019 Update 5
LogitBoost algorithm parameters. More...
| Parameter | ( | ) |
Default constructor
| Parameter | ( | const services::SharedPtr< weak_learner::training::Batch > & | wlTrainForParameter, |
| const services::SharedPtr< weak_learner::prediction::Batch > & | wlPredictForParameter, | ||
| double | acc = 0.0, |
||
| size_t | maxIter = 10, |
||
| size_t | nC = 0, |
||
| double | wThr = 1e-10, |
||
| double | zThr = 1e-10 |
||
| ) |
Constructs LogitBoost parameter structure
| [in] | wlTrainForParameter | Pointer to the training algorithm of the weak learner |
| [in] | wlPredictForParameter | Pointer to the prediction algorithm of the weak learner |
| [in] | acc | Accuracy of the LogitBoost training algorithm |
| [in] | maxIter | Maximal number of terms in additive regression |
| [in] | nC | Number of classes in the training data set |
| [in] | wThr | Threshold to avoid degenerate cases when calculating weights W |
| [in] | zThr | Threshold to avoid degenerate cases when calculating responses Z |
| double accuracyThreshold |
Accuracy of the LogitBoost training algorithm
| size_t maxIterations |
Maximal number of terms in additive regression
| size_t nClasses |
Number of classes
| double responsesDegenerateCasesThreshold |
Threshold to avoid degenerate cases when calculating responses Z
| double weightsDegenerateCasesThreshold |
Threshold to avoid degenerate cases when calculating weights W
For more complete information about compiler optimizations, see our Optimization Notice.