Default ARTMAP Neural Networks for Classification of Anthrax Time Series from Inhalation Anthrax Models

摘要

The possibility of the usage of deadly aerosolized pathogens, particularly anthrax, in bioterrorist attack has raised tremendous concerns in recent years. Several anthrax incubation models have been introduced in order to characterize the incubation period of human inhalation anthrax. It is important to accurately identify the model that fits best with the observed anthrax time series, which directly affects the prediction results of the severity of the potential anthrax attacks. Here, we applied Default ARTMAP, an important neural network algorithm for classification, to separate anthrax time series generated from different inhalation anthrax models. Experimental results on anthrax time series derived from major inhalation anthrax models, together with anti-patterns and a smallpox time series, demonstrate the effectiveness of Default ARTMAP in identifying anthrax time series derived from different models, as well as discriminating unrelated cases.