Hello all,though I cannot comment specifically on the case of MH370, I can say a little bit about bayesian methods in such situations. This is mainly because I supervised a student's talk about Bayesian Search Theory [7]; I'm by no means an expert on this stuff.
The application of bayesian methods to rescue operations has quite a long tradition, belonging to the Operations Research area. One of the earliest documented cases is the search for the USS Scorpion [3]. Besides a good model for the true location of the missing object, it is also of concern to schedule the search in such a way to find the object as early as possible, and without using too much resources (ships, planes). E.g. one would not look at a location early that hampers search (high chance of false negative result), even if the probability of the object being located there is high. So we are dealing with a planning problem (POMDP).
During the search for Scorpion, MCMC simulations were used for inference. I guess that the models currently used for the MH370 event are of the kind you can treat with tools like BUGS and JAGS.
Some days ago there was also an article on bbc.co.uk [5] about how bayesian methods are used to deal with such situations. The article is interesting to read, and also refers to how statisticians finally managed to help in finding flight Air France 447 [6] two years after it crashed. The Wikipedia article on 447 also contains a link to "Bayesian Search Theory" [7], which might be of interest.
Hope this was helpful, Thomas # References (also some unreferenced ones) [1] Koopman, B.O., The Theory of Search: II Target Detection Operations Research, 4(5):503–531, 1956 [2] Stone, J.D., Theory of Optimal Search, Academic Press, 1975 [3] Richardson, H.R. and Stone, L.D., Operations Analysis During the Underwater Search for Scorpion, Naval Res. Logist. Quart., 18 (1971) 141–157 [4] Benkoski, S.J. and Monticino, M.G. and Weisinger, J.R., A Survey of the Search Theory Literature, Naval Research Logistics, 38(4):469—494, 1991 [5] http://www.bbc.com/news/magazine-26680633 [6] Wiki on AF 447: http://en.wikipedia.org/wiki/Air_France_Flight_447[7] Wiki on Bayesian Search Theory: http://en.wikipedia.org/wiki/Bayesian_search_theory
On 21.03.2014 00:01, Sajjad Haider wrote:
Hi, Has some work been done (which is accessible via public domain) to find the best explanation of MH370 disappearance using a Bayesian network? It might be easier to develop a BN once that ill-fated plane is found but it would be nice to see if the community can develop a BN to explain how Bayesian approach would solve this extremely uncertain situation. Secondly, is there any public space available where people can collectively build a BN and discuss/argue over its structure and parameters? Best regards, Sajjad _______________________________________________ uai mailing list uai@ENGR.ORST.EDU https://secure.engr.oregonstate.edu/mailman/listinfo/uai
-- Dipl.-Inf. Thomas Geier Universität Ulm Faculty of Engineering and Computer Sciences Institute of Artificial Intelligence James-Franck-Ring D-89081 Ulm Phone: +49 (0)731 50 - 24272 Fax: +49 (0)731 50 - 24119 Homepage: http://www.uni-ulm.de/in/ki/staff/thomas-geier.html
<<attachment: thomas_geier.vcf>>
_______________________________________________ uai mailing list uai@ENGR.ORST.EDU https://secure.engr.oregonstate.edu/mailman/listinfo/uai
