Machine Learning-Based Intrusion Detection System for Cyber Attacks in Private and Public Organizations

Abstract

Cyber-attacks have proven to be a force for hacking groups and state-sponsored organizations seeking to level the playing field with competitors. The hacker threat paired with the enormously hazardous and costly danger of fraud or intellectual property theft by insiders has created a volatile situation in private and public organizations. While a majority of internal breaches are due to employee negligence or human error, attacks by malicious insiders with access to sensitive company information have increased dramatically in recent years. Threats of financial loss, theft of sensitive information, and destruction to critical sectors have made cybersecurity a top security priority around the globe. Whereas the increase in frequency and complexity of attacks on the industry has increased the danger of being unprepared, it also has influenced the cost of preventing and recovering from cyber-attacks. To construct a machine learning bases instruction detection system is capable of detecting Cyber-attacks in the private and public sectors in Nigeria and the whole world. The results show that Random Forest and Random Tree algorithms outperform the other algorithms in their level of precision and F-measure as they are above 99% and 98% respectively, while the Random Forest outperforms the others by its detection rate. However, the Random Forest and Random Tree algorithms are more efficient in performing classification exercise on the Test datasets