MATLAB Bayesian Network Toolkit (BNT) - Bayesian Learning and Inference Framework

This is a MATLAB-based Bayesian Network Toolkit (BNT) that offers a rich set of low-level functions, including comprehensive Bayesian learning algorithms and Bayesian inference procedures. The toolkit supports multiple inference methods such as junction tree algorithm and approximate sampling techniques, along with parameter learning capabilities using EM algorithm and structure learning through score-based methods. You can leverage this toolkit to implement various Bayesian network classifiers, including Naive Bayes, Tree-Augmented Naive Bayes (TAN), and general Bayesian classifiers, facilitating complex data analysis and prediction tasks. The package provides detailed implementations covering all aspects of Bayesian network applications, featuring object-oriented design with classes for nodes, graphs, and inference engines. Whether you're a beginner or professional, this toolkit meets diverse requirements through its powerful functionality and flexible operation. Key functions include graph construction using 'mk_bnet', inference execution via 'jtree_inf_engine', and parameter learning with 'learn_params'. You can utilize this toolkit to deeply study Bayesian network principles and applications, providing strong support and guidance for your projects and research through its well-documented API and modular code structure.