Learning Future Terrorist Targets using Attention Based Hybrid CNN and BiLSTM Model

Abstract

Terrorism is complex, with a huge scope of belief systems, reasons, entertainers, and objectives, and it represents a danger not exclusively to assemblies and organizations, yet in addition to humankind all in all. In this manner, concentrating on people or regions at high threat of being designated can support the improvement of precaution measures and the distribution of assets to protect these objectives. Utilizing true information on assaults that happened in South Asia from 2009 to 2019, our undertaking proposes the utilization of deep learning to map the associations among terrorist attacks, capturing their functional similarities and conditions. It will be used to determine what target districts that are at the most risk of being picked next. The execution will include LSTM, Bi-LSTM, CNN, CNN-LSTM models. The project emphasises on a CNN-BiLSTM model that is improved using attention layers, hence called the CNN-BiLSTM Attention Mechanism.