#include <Classifier.hpp>
Detailed Description
- Classifier class is used as a base class for binary classifier objects (e.g., NCC, FLDA+NCC, SVM, AdaBoost).
- Classifier object attributes are public for convenience.
- Although the public type class attributes are easier to handle, this also entails certain pitfalls. The user should manipulate these objects with extra caution.
- Attributes:
- plus : Vector of integers used to hold the initial class labels that the binary classifier labels with +1 (e.g., when using Discriminant Error Correcting Output Codes, although the classifier is binary, it might be used to discriminate a binary decomposition of a multiclass problem. In this case the "plus" vector will hold the initial class labels, of the classes that belong to the bipartition which is marked with +1 in the created coding matrix).
- minus : Vector of integers used to hold the initial class labels that the binary classifier labels with -1 (e.g., when using Discriminant Error Correcting Output Codes, although the classifier is binary, it might be used to discriminate a binary decomposition of a multiclass problem. In this case the "plus" vector will hold the initial class labels, of the classes that belong to the bipartition which is marked with +1 in the created coding matrix).
- training_error : Training error attained by classifier on a binary specific problem.
Member Function Documentation
| virtual double Classifier::predict |
( |
const rowvec & |
t | ) |
const [pure virtual] |
Field Documentation
The documentation for this class was generated from the following file:
1.7.4
