CW learning algorithm.
CW - Confidence Weighted Learning
Reference: Dredze, M., Crammer, K., Pereira, F.: Confidence-weighted linear classification. In: Proceedings of the 25th International Conference on Machine Learning, pp. 264-271. ACM (2008) http://webee.technion.ac.il/people/koby/publications/icml08_variance.pdf
Documentation
algorithms/icl_learn_CW
ILS = icl_learn_CW(ILS, x, y, yp, mode)
Inputs:
ILS | structure describing the learning system |
x | double input vector in [-10,10] |
y | true label, i.e. target value of the example, in [-1, 1] for regression or [-1; 1] for classification |
yp | predicted label, i.e. current output of the approximation, in [-1, 1] for regression or [-1; 1] for classification |
mode | operating mode (regression=1, classification=2) |
Outputs:
ILS | updated matlab structure describing the learning system |
Calls:
Called by:
- none
Authors:
- Simon Herkenhoff (sherkenh(at)uos.de)
- Andreas Buschermöhle (andbusch(at)uos.de)
Last change: 2013-07-13 - Version 1.5