Jovanovic, L.; Spalević, Ž.; Bacanin, N.; Simić, M.; Marković, F. Insider Threat Detection Form Irregular Login Patterns with Metaheuristic Optimized AdaBoost. Preprints2024, 2024091500. https://doi.org/10.20944/preprints202409.1500.v1
APA Style
Jovanovic, L., Spalević, Ž., Bacanin, N., Simić, M., & Marković, F. (2024). Insider Threat Detection Form Irregular Login Patterns with Metaheuristic Optimized AdaBoost. Preprints. https://doi.org/10.20944/preprints202409.1500.v1
Chicago/Turabian Style
Jovanovic, L., Milan Simić and Filip Marković. 2024 "Insider Threat Detection Form Irregular Login Patterns with Metaheuristic Optimized AdaBoost" Preprints. https://doi.org/10.20944/preprints202409.1500.v1
Abstract
This paper addresses a critical concern in intrusion detection within the broader realm of cyber security, particularly focusing on login activity involving the majority of normal users. Utilizing the AdaBoost classifier, the study employs various optimizers to enhance performance by selecting optimal control parameters. A specially tailored version of Crayfish Optimization Algorithm (COA) is introduced to cater to the unique requirements of this investigation. Through a comparative analysis of a simulated publicly available dataset, models optimized by the modified algorithm demonstrate superior outcomes, achieving an accuracy of 94.6128% and displaying an adaptive convergence rate capable of navigating local minima to identify optimal solutions. The best-performing model undergoes SHapley Additive exPlanations (SHAP) analysis to identify key contributing features. Limitations arise from the computational intensity of the optimization process, necessitating consideration of limited populations and smaller numbers of estimators during simulations. Future endeavors will extend the methodology to incorporate additional user actions in classification, with a focus on addressing computational constraints as hardware advancements occur. Proposed, modified algorithm could be applied to deal with various optimization tasks, beyond the scope of this study.
Computer Science and Mathematics, Artificial Intelligence and Machine Learning
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.