IDSAI Research Seminar with Dr Livio Fenga, Senior Lecturer, University of Exeter Business School

Terrorism represents a multifaceted political and social phenomenon, posing a substantial threat to the safety and security of the international community. It has emerged as a major impediment to the sustainable development of global social security. The fight against terrorism  is integral to global security governance, addressing a sustainability concern crucial for ensuring the continuous development of global security. Predicting terrorist attacks with a high degree of accuracy is an extremely challenging task due to the complex and dynamic nature of terrorism. Terrorist organizations often operate in clandestine and decentralized ways and their activities are carried out by a diverse range of groups and individuals with varying motivations, and the factors that drive these actions are often complex and difficult to anticipate. Some attacks, like lone-wolf or small-scale incidents, can be particularly difficult to predict as they may be impulsive and not part of a broader plot. Whilst there is no known foolproof method to anticipate terrorist actions, indications on their predictability can be drawn within the framework of stochastic processes and multiresolution analysis. In this regard, a wavelets—based approach for the decomposition of the data at different frequency bands will be illustrated.   The forecasting method proposed is based on a computational intelligence method of the type Extreme Learning Machine, used in conjunction with a nonlinear model belonging to the class SETAR (Self-Exciting Threshold AutoRegressive). Standard ARIMA (AutoRegressive Integrated Moving Average) models will also be used to model linear dynamics. Finally, the data employed, spanning from 1970 to 2020, are released on a regular basis by the Global Terrorism Database — which is publicly and freely available at the web address: https://www.start.umd.edu/gtd/ — managed by the National Consortium for the Study of Terrorism and Responses to Terrorism (START). This dataset is widely recognized as the most comprehensive database for documenting global terrorist activity.