Multifractal Singularity Spectrum for Cognitive Cyber Defence in Internet Time Series

2. Cognitive Computing And Complexity

In the last few years, cognitive computing has emerged as a promising domain and has shown capabilities to outperform the traditional artificial intelligence tools including machine learning and neural networks. Cognitive computing is a combination of the techniques and tools which human use to infer and analyze objects. These include human decision making, mental inference, brain sciences and reasoning capabilities. All the traditional tools to analyze, predict and infer knowledge including but not limited to probability, statistics, dynamical systems, signal processing, machine learning, multiscale analysis and multifractal analysis are being used to define different aspects of cognitive computing. Primary goal of cognitive computing is to model improvements in making decision using experience from historical data (learning) and stimulus-response mechanism of human behavior to adapt according to the changing context and environments (adaptability). As opposed to traditional decision making computing systems and tools which aim to reduce complexity, cognitive computing aims to utilize the concepts of complexity and uncertainty to achieve tractability and robustness (Brasil, L. M., Mendes de Azevedo, F., Barreto, J. M., & Noirhomme-Fraiture, M.) (Kinsner, W. (2009, July-September)).