It is well known that many hard tasks considered in machine learning and data mining can be solved in a rather simple and robust way with an instanceand distance-based approach. In...
In active learning, a machine learning algorithm is given an unlabeled set of examples U, and is allowed to request labels for a relatively small subset of U to use for training. ...
Support vector machines (SVMs) excel at two-class discriminative learning problems. They often outperform generative classifiers, especially those that use inaccurate generative m...
This paper proposes a novel Mass Spectrometry data profiling method for ovarian cancer detection based on negative correlation learning (NCL). A modified Smoothed Nonlinear Energy ...