Project Description

Probabilistic graphical models (PGMs) allow the representation of probability distributions with many variables in a compact form, also they help to make probabilistic inference (estimate the probability of certain variables given other known) efficiently.

PGMs include: Bayesian classifiers, Bayesian networks, Markov random fields, etc.

PGM´s have many applications in medicine, expert systems, industrial diagnosis, image analysis, robotics and many others.

This models has been extended, on one hand, to dynamic processes, including hidden Markov models and dynamic Bayesian networks. On the other hand, there are models that include decisions and utilities, such as dynamic influence diagrams (DIDs) and Markov decision processes (MDPs).

The DyNaMo project will concentrate its efforts on the probabilistic dynamic graphical models and its applications on medicine and industry.

The main objectives are:

1. To contribute in the development of dynamic probabilistic graphical models (DPGMs), in particular in techniques for building this type of models, in more efficient inference techniques, and in explanation methodologies. All these developments will be implemented in software tools that will be the basis to develop applications.
2. To apply DPGMs to the solution of problems in Medicine, in particular to analyze the cost-effectiveness of the vaccine for the human papilloma virus, in the rehabilitation of people that have suffered a stroke, prognosis and treatment of ovary cancer and the selection of drugs again VIH .
3. To apply DPGMs in the solution of complex problems in industry, such as the optimal control of industrial processes considering security, efficiency, and the impact in the environment; as well as in training the operators of this type of systems.