An efficient trustworthy cyberattack defence mechanism system for self guided federated learning framework using...

Area of Science:

• Cybersecurity
• Machine Learning
• Decentralized Systems

Background:

• Conventional centralized threat intelligence models struggle with advanced cyberattacks.
• Federated learning (FL) offers a privacy-preserving, decentralized approach to cybersecurity.
• Machine learning (ML) and Deep Learning (DL) advancements are crucial for robust cyberattack defense.

Purpose of the Study:

• To present a novel Cyberattack Defence Mechanism System for Federated Learning Framework (CDMFL-AIDCNN).
• To enhance cybersecurity resilience and privacy in distributed systems using FL and attention-based deep learning.
• To improve cyberattack detection accuracy and efficiency.

Main Methods:

• Utilized Z-score normalization for data preprocessing.
• Employed Dung Beetle Optimization (DBO) for effective feature selection.
• Developed an Attention Induced Deep Convolution Neural Networks (AIDCNN) model, integrating convolutional neural networks, bidirectional long short-term memory, gated recurrent units, and attention (CBLG-A), optimized by the Growth Optimizer (GO).

Main Results:

• The CDMFL-AIDCNN technique demonstrated superior performance in cyberattack detection.
• Achieved high accuracy rates of 99.07% on the CIC-IDS-2017 dataset and 98.64% on the UNSW-NB15 dataset.
• Validated the effectiveness of FL and deep learning fusion for enhanced cybersecurity.

Conclusions:

• The proposed CDMFL-AIDCNN system significantly improves cyberattack defence mechanisms in federated learning frameworks.
• The integration of FL with advanced deep learning models offers a robust and privacy-preserving solution for modern cybersecurity challenges.
• This approach paves the way for more adaptive and resilient cybersecurity practices in distributed environments.