Reconciling Privacy and Security: A Multi-Objective Optimization Framework for Ethical Cyber Defense Systems (MOOF-ECDS)
ID：91
Submission ID：488 View Protection：ATTENDEE
Updated Time：2026-07-22 16:09:52 Hits：18
Online
Start Time：2026-07-30 16:10 （Asia/Kolkata）
Duration：15min
Session：[S3] Cyber Security » [S3-2] Cyber Security
Presentation File
Attachment File
Tips: The file permissions under this presentation are only for participants. You have not logged in yet and cannot view it temporarily.
Abstract
The increasing complexity of cyber threats requires correspondingly improved security systems, frequently utilizing deep learning (DL) for intrusion detection. Nonetheless, these potent systems often function with excessive permissions, resulting in considerable privacy violations due to the over-collection and examination of network and user data. This issue establishes a significant ethical and technical dichotomy between security effectiveness and privacy protection. This research introduces an innovative ethical cyber defense framework that systematically conceptualizes this issue as a multi-objective optimization problem (MOOP). Our approach employs a hybrid metaheuristic algorithm, combining the exploratory power of the Grey Wolf Optimizer (GWO) with the predictive accuracy of a Deep Neural Network (DNN) to dynamically tune the parameters of a network intrusion detection system (NIDS). The main goals are to concurrently enhance threat detection rates (true positive rate) and reduce privacy-invasive data collecting, measured by an innovative privacy impact score. The GWO-DNN hybrid is trained and validated on the CIC-IDS2017 dataset, augmented with synthetic data to simulate privacy-sensitive scenarios. Experimental findings indicate that the proposed framework attains Pareto-optimal equilibrium, substantially decreasing the privacy footprint by as much as 40% relative to a security-maximized baseline, while preserving a high detection accuracy of 98.5%. The paper indicates that multi-objective optimization offers a mathematically rigorous approach to creating cybersecurity systems that are both secure and fundamentally ethical, adhering to contemporary data protection requirements such as GDPR and CCPA.
Keywords
Cybersecurity, Deep Learning, Ethical AI, Intrusion Detection Systems, Grey Wolf Optimizer, Multi-Objective Optimization, Privacy Preservation.
Speaker
Comment submit