Individual machine learning algorithms, such as inductive learning techniques,
perform with varying accuracy on different data sets. It is impossible to
determine a priori which algorithm will perform better on what data set,
based on the data set's characteristics, e.g. missing values, noise, number
of attributes, type of attributes, etc.
SmartXAutofill has the solution:
Our technology incorporates many classifiers that have been trained on the same data set by a variety of machine learning algorithms into a single classifier, an ensemble classifier.
Ensemble learning is a machine learning technique that selects a collection, or ensemble, of hypotheses from the hypothesis space and combines their predictions. An ensemble consists of a set of independently trained classifiers whose predictions are combined when classifying new instances.
Various classifiers are integrated in our ensemble classifier. It uses a voting mechanism to determine which of the classifiers to believe. The ensemble rewards classifiers that make the correct prediction and punishes those that are wrong. After a training period the ensemble learns to prefer the best classifier for a specific data set.
SmartXAutofill is a patented technology .
Copyright © 2010 by the University of Kansas
Web Design and Technology Services by:
Information and Telecommunication Technology Center