Explainable AI‐Driven Firewall Evaluation: Empowering Cybersecurity Decision‐Making for Optimal Network Defense DOI

Zaheen Fatima,

R. Jahir Hussain,

Azhar Dilshad

и другие.

Security and Privacy, Год журнала: 2025, Номер 8(3)

Опубликована: Апрель 4, 2025

ABSTRACT The probability of network attacks is increasing daily due to the continuous development tools and techniques that bypass firewall other security boundaries. This motivates researcher towards upgradation advancement in adaptive Artificial Intelligence (AI) based intrusion detection systems (IDS). traditional machine learning (ML) IDS has its limitations noise unexplainable nature decision‐making took place during implementation ML algorithm for categorization attacked or normal data packets arrive network. To address these issues, this research proposes an with understanding decisions through Explainable artificial intelligence (XAI). dataset used experimentation IoTID20, which extracts a real‐time Internet Things (IoT) environment. paper compares accuracy results different algorithms such as Random Forest (RF), Extreme Gradient Boosting (XGBoost), Multilayer Perceptron –Neural Networks (MLP‐NN), Deep Neural (DN), Decision Tree (DT). During classification attacks, shows 93.80%, XGBoost 97.30%, 99.99%, MLP Classifier—Neural Network 95.90%, (DNN) 94.60%. These are also analyzed Precision, Recall, F1‐Score. proposed method incorporating XAI increases automation process high explainability decision categorize anomalous IoT remarkable achievement provides better intuitions good protection resistance novel unknown attacks.

Язык: Английский

Explainable AI‐Driven Firewall Evaluation: Empowering Cybersecurity Decision‐Making for Optimal Network Defense DOI

Zaheen Fatima,

R. Jahir Hussain,

Azhar Dilshad

и другие.

Security and Privacy, Год журнала: 2025, Номер 8(3)

Опубликована: Апрель 4, 2025

ABSTRACT The probability of network attacks is increasing daily due to the continuous development tools and techniques that bypass firewall other security boundaries. This motivates researcher towards upgradation advancement in adaptive Artificial Intelligence (AI) based intrusion detection systems (IDS). traditional machine learning (ML) IDS has its limitations noise unexplainable nature decision‐making took place during implementation ML algorithm for categorization attacked or normal data packets arrive network. To address these issues, this research proposes an with understanding decisions through Explainable artificial intelligence (XAI). dataset used experimentation IoTID20, which extracts a real‐time Internet Things (IoT) environment. paper compares accuracy results different algorithms such as Random Forest (RF), Extreme Gradient Boosting (XGBoost), Multilayer Perceptron –Neural Networks (MLP‐NN), Deep Neural (DN), Decision Tree (DT). During classification attacks, shows 93.80%, XGBoost 97.30%, 99.99%, MLP Classifier—Neural Network 95.90%, (DNN) 94.60%. These are also analyzed Precision, Recall, F1‐Score. proposed method incorporating XAI increases automation process high explainability decision categorize anomalous IoT remarkable achievement provides better intuitions good protection resistance novel unknown attacks.

Язык: Английский

Процитировано

0