Analyze and Forecast the Cyber Attack Detection Process using Machine Learning Techniques
Authors:
- 6 July 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1732-1738
Abstract
One of the most crucial global concerns is the issue of cybercrime, which leads to significant financial losses for nations and their citizens every day. The frequency of cyberattacks has steadily increased, emphasizing the need to identify the individuals behind these criminal activities and understand their strategies. Detecting and preventing cyberattacks pose significant challenges, but recent advancements have introduced security models and prediction tools based on artificial intelligence to tackle these issues. Although there is a wealth of literature on crime prediction strategies, they may need to be more effectively suited for awaiting cybercrime and cyber-attack techniques. One potential solution to address this problem involves utilizing real-world data to determine the occurrence of an attack and identify the responsible party. This information encompasses details about the offense, offender demographics, property damage, and attack vectors. Forensic teams can collect information from victims of cyber-attacks through application processes. This research study employs machine learning techniques to analyze cybercrime using two models and predict how the attributes can contribute to identifying the method of cyber-attack and the criminal. This study has compared eight different machine-learning techniques, and discovered that they yielded similar results in terms of accuracy. The Support Vector Machine (SVM) linear model achieved the highest accuracy rate among the various cyber-attack methods tested. In the first model, valuable insights on the types of attacks victims were likely to face. Logistic regression, with a high success rate, was the most effective strategy for identifying malicious actors. The second model focused on comparing offender and victim attributes to make predictions regarding identification. Our findings indicate that the likelihood of becoming a victim of cyberattacks decreases with higher levels of education and wealth. This proposed concept is eagerly estimated for implementation by cybercrime departments, as it will simplify the detection of cyber-attacks and enhance the efficiency of the battle against them.Keywords
This publication has 22 references indexed in Scilit:
- A hybrid approach estimates the real-time health state of a bearing by accelerated degradation tests, Machine learning2021
- Cyber-attack method and perpetrator prediction using machine learning algorithmsPeerJ Computer Science, 2021
- Thwarting Cyber Crime and Phishing Attacks with Machine Learning: A Study2021
- Network Attacks Detection Methods Based on Deep Learning Techniques: A SurveySecurity and Communication Networks, 2020
- Conceptualisation of Cyberattack prediction with deep learningCybersecurity, 2020
- An Ensemble of Deep Recurrent Neural Networks for Detecting IoT Cyber Attacks Using Network TrafficIEEE Internet of Things Journal, 2020
- An Ensemble Deep Learning-Based Cyber-Attack Detection in Industrial Control SystemIEEE Access, 2020
- Detecting Port Scan Attempts with Comparative Analysis of Deep Learning and Support Vector Machine Algorithms2018
- A Study On Various Cyber Attacks And A Proposed Intelligent System For Monitoring Such Attacks2018
- Performance Comparison of Machine Learning Models for DDoS Attacks Detection2018