Cyber threats have become increasingly dynamic and complex due to the fusion of technologies, networks and systems. As a corollary the artificial intelligence techniques have become the focal point for the cyber security research, as they are viewed to be more suited to tackling the modern cyber threats. Specifically the generalization performance capability of neural networks enables them to lend themselves to a whole gamut of cyber threats including emerging threats. This is different from domain specific techniques whose customized function restrict them to a specific type of threat, hence increasing the risk of threats going undetected. In this paper, a neural network ensemble for malware detection is proposed. The approach is based on a hybrid search mechanism where the optimizing of individual networks is done by an adaptive memetic algorithm with Tabu Search, which are also used to improve hidden neurons and weights of neural networks. The adaptive memetic algorithm combines global and local search optimization techniques to overcome premature convergence and obtain an optimal search outcome. The results from the empirical study prove that the proposed method is more adaptive and efficient at detecting a range of cyber threats, as it generates better results than the existing methods.

Akandwanaho and Kooblal