Poison Identification Based on Bayesian Method in Biochemical Terrorism Attacks

摘要

This paper intends to provide help for poison identification in biochemical terrorism attacks according to the observed preliminary symptoms of the poisoning people. We find the optimal initial parameters for two Bayesian network structure learning algorithms, Hill-climbing algorithm and K2 algorithm. Bootstrap data expansion and Gibbs data correction combining with tree augmented naive Bayesian network (TAN-BN) are used to expand the original small data set to improve the learning effect of these algorithms. We find the best combination of learning Bayesian network structure for our data set with the characteristic of containing only confirmed cases. Finally we use the Bayesian network learned from the group of anthrax infection data to analyze the relation between anthrax and its poisoning symptoms and make inference to get the probability of anthrax class node. This method can be extended to a variety of biochemical reagents, and the result of the inference can be used to guide emergency rescue after certain biochemical terrorism attack.