A hybrid approach for efficient anomaly detection using metaheuristic methods.

Area of Science:

• Computer Science
• Cybersecurity
• Artificial Intelligence

Background:

• Network intrusion detection systems (NIDS) are crucial for cybersecurity.
• Anomaly detection techniques are widely used in NIDS.
• Metaheuristic methods have been applied for anomaly detector generation, but multi-start approaches remain underexplored.

Purpose of the Study:

• To propose a novel hybrid approach for anomaly detection in large-scale datasets.
• To investigate the efficacy of multi-start metaheuristic methods combined with genetic algorithms for generating anomaly detectors.
• To address the gap in literature regarding the use of multi-start metaheuristics in NIDS.

Main Methods:

• A hybrid approach combining multi-start metaheuristic methods and genetic algorithms for anomaly detector generation.
• Inspiration drawn from negative selection-based detector generation principles.
• Evaluation performed on the NSL-KDD dataset, a derivative of KDD CUP 99.

Main Results:

• The proposed approach successfully generates a suitable number of effective anomaly detectors.
• Achieved a high accuracy of 96.1% in detecting network intrusions.
• Demonstrated superior performance compared to other machine learning algorithms on the NSL-KDD dataset.

Conclusions:

• The hybrid multi-start metaheuristic and genetic algorithm approach is effective for network intrusion detection.
• This method offers a promising solution for anomaly detection in large-scale network environments.
• The study highlights the potential of underutilized metaheuristic techniques in enhancing cybersecurity.