| SNN-4: Software for Graphical Models in Decision Making Systems |
|
Organization responsible: SNN Adaptive Intelligence People involved:
Project description: This project aims to develop a software library and an interface for modelling, inferencing, decision making and control in graphical models for single agent and collaborative multi-agent systems. In project SNN-1, SNN-2 and SNN-3 methods and algorithms on graphical models for multi-agent systems are being investigated and further developed. In this subproject, we aim to deliver general purpose software that implements these methods and algorithms. Starting point will be software for learning and inference in probabilistic models that we have developed in previous projects. One of these packages is BayesBuilder, which is a library and a graphical user interface for modelling and reasoning in static Bayesian nets. In this project we will aim to extend our software to decision support and multi-agent systems. The software is aimed to do modelling and simulations, for instance it should be able to optimise decisions in what-if scenarios in decision support systems. This project aims to develop a software library and an interface for modelling, inferencing, decision support, decision making and control in graphical models for single agent and collaborative multi-agent systems. Our current BayesBuilder software consists of C++ and JAVA routines. Both an API and a graphical user interface are implemented. Furthermore we have developed an open-source library with approximate inference methods (libDAI). Our approach is to improve this software and extend it. Publications: |