Software

Bayesia, a company created by researchers specialized in Bayesian Networks and Machine Learning, launches BayesiaLab, a powerful laboratory for mining your data with Bayesian Networks.

Briefly, BayesiaLab offers a broad set of Bayesian Networks learning methods, from learning parameters with Bayesian estimations of the probabilities, up to learning the network structure. Three types of structural learning are available:

• A set of very effective unsupervised learning methods for discovering the whole set of probabilistic relationships existing within the data;
• Unsupervised learning for new concepts discovery (segmentation, clustering);
• Various supervised learning methods focussing on a target variable (from classical Naive Bayes classifiers to an original Augmented Markov Blanket learning allowing a tremendous gain in time and models readability).

In order to improve the Bayesian networks readibility, BayesiaLab also comes with a complete graphical analysis toolbox that allows discovering very quickly the strength and the type of the probabilistic relations encoded by the (learned) models.

For more information and download an evaluation version, see http://www.bayesia.com.