Classification of Cognitive Patterns of Hackers Using Machine Learning

Abstract

Nowadays, the importance of computer security has risen to unprecedented levels; in addition to protecting digital assets, it is also necessary to safeguard the privacy of our financial institutions, companies, education, and defense, among others, from recurrent, sophisticated, and constantly evolving cyber threats. For this, it is necessary to combine different methodologies, techniques, and computer security tools; among these, we use Honeypots, Machine Learning, and ELK Stack. In addition, analyzing the psychology of the hacker and knowing how he thinks and behaves provides us with an advantage to counter them. In the present research, an immersion is made in two fields, such as the use of honeypots in computer security and the analysis of psychology, that is, what are their motivations or interests, and also, the instruments used to measure all the above mentioned. Afterward, attack data was collected using the T-Pot Honeypot, and the Big Five Personality Traits instrument was applied. Subsequently, a database was generated with all this information, which was used for the analysis through Machine Learning algorithms and neural networks with confusion matrices composed of prediction and real data. As for the classification of cognitive patterns acquired through Honeypots and ML algorithms for processing, it is a new field that provides valuable information to understand better how cyber attackers or hackers operate and develop more effective countermeasures. It is necessary to develop tools (psychological tests) targeted at hackers to have better results in future research. ML algorithms such as Neural Networks using a sequential model and Random Forest using 150 predictors adequately fit the training and test data.