According to the article “Validation of Reliability Computational models using Bayes Networks”, Bayes can be used to test certain models if the full-scale testing is impossible. There exist several studies which pay attention to the concepts and terminology that draw a conclusion about the validity of large-scale computational models, such as the Advanced Simulation and Computing Initiative (ASCI) program of the United States Department of Energy, American Institute of Aeronautics and Astronautics, American Society of Mechanical Engineers Standards Committee as well as the Defense Modeling and Simulation Office Model. The models reliability is analyzed through the demand vs. capacity format, related to their desired performance criterion. Assuming that R stands for the capacity and S for the demand, the performance function G (r,s) = R-S fails if G, standing for the performance, gets negative. In order to analyze the reliability of models the Bayesian hypotheses testing is used to compare the prior density values and the posterior distributions. The metric is known as the Bayes factor, which contains the “ration of posterior and prior density values at the predicted value of G corresponding to a given set of input data”. The same input data is used in the model and in the validation experiment, so that the predicted output can be compared with the measured output in the end. Computing the Bayes factor metric, we measure the difference between the model prediction and the experimental data. A difference in the outcome may occur through numerical errors which are related to solution convergence and model resolution, stochastic analysis errors, and measurement errors in the model input.

In general we can say that we compare the prediction and the observation to come up with a validation of our model. Using the Bayes model we have to update our data frequently each round. The article comes to the conclusion that “The Bayes network is a causal network, used as an inference engine for the calculation of beliefs or probability of events given the observation/evidence of other events in the same network.” That is why Bayes model is applied in many fields, such as artificial intelligence, engineering decision strategy, safety assessment of software-based systems, and model-based adaptive control.

source:

http://www.sciencedirect.com.libproxy.temple.edu/science/article/pii/S0951832004001103

### Like this:

Like Loading...

*Related*

It makes a lot of sense that Bayes’ model is being used in so many different fields, especially science based. For experiments like the ones in NASA or ASCI program, I think Bayes’ model is the most useful. Of course all known information must be considered, especially when there are thousands of potential outcomes and every bit of knowledge will help narrow the outcomes to the ones we want. Also, because Science has always been (to me) a field in which exploration and research is fundamental, Bayes model fits right in.