<img height="1" width="1" alt="" style="display:none" src="https://www.facebook.com/tr?id=648880075207035&amp;ev=NoScript">
BayesiaLab

The BayesiaLab Digest - February 9, 2015
Asymmetric Threat Detection with Bayesian Networks

Here is today's citation of new and interesting applied research with Bayesian networks:

Dragos, Valentina, Juergen Ziegler, and Paulo C. Costa.
Description and Assessment of a User Oriented Approach for Asymmetric Threat Detection.
Technical Report. George Mason University, 2013. 
http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA606212.

threat-detection

Abstract
Asymmetric threats pose a difficult challenge to situational awareness systems. Current approaches for predicting or even detecting an asymmetric threat rely heavily on human knowledge, creating scalability issues due to the vast amount of data to be analyzed. Attempts to automate this process require a combination of advanced knowledge representation techniques to capture what human experts know about the domain and inferential reasoning approaches capable to work with incomplete, uncertain data. In our current research, we apply a verb-oriented ontology to capture actions, features, indicators, and other domain elements that are relevant to asymmetric threat detection. Then, these elements are input to a Bayesian network that will calculate the posterior probability of a threat given the input. As in any complex process, evaluation is a key asset for ensuring that nothing is neglected and partial results are consistent with the expectations. This paper describes our approach for asymmetric threat detection and emphasizes how we are leveraging the Uncertainty Representation and Reasoning Evaluation framework (URREF), to support its evaluation. We discuss how the sources of uncertainty are identified and how we assess its impact to the outcome of the detection system.

2017 BayesiaLab Conference