Abstract :

With The Growing Sophistication And Prevalence Of Cyber Threats, An Intelligent And Proactive Security Approach Is Required To Accurately Detect And Predict Cyber Attacks. This Study Introduces A Complex AI-driven Cyber Attack Prediction Framework Using The CICIDS2017 Dataset, Which Integrates ML, DL, GAI, And XAI Techniques. The Preparation Of The Data Is Comprehensive, Such As Removing Missing Values, Duplicate Data, Encoding Labels, Data Normalisation And Dimension Reduction Through PCA, To Help Improve The Learning Efficiency. We Train And Test Multiple ML Classifiers Including DT, RF, ET Classifier, Logistic Regression, Gaussian Naive Bays, And A Voting Classifier Of Combination Of RF, Light Gbm And Xgboost As Well As Multiple DL Models Including DNN, CNN, LSTM, CNN–LSTM And CNN–LSTM–GRU. GAE, GAN And DistilGPT2 Models Are Used To Generate Synthetic Attack Patterns And Improve Anomaly Representation. The Experimental Results Show That The Voting Classifier Achieved The Maximum Accuracy Of 99.6%, While LSTM Achieved 99.3%, Indicating That The VC And LSTM Model Had Excellent Detection Accuracy For Various Attacks Such As DoS, DDoS, PortScan, Bot And Infiltration Attacks. For Model Interpretability Use LIME And SHAP. The Framework Is Developed In Flask Which Takes Care Of Authentication, Input Processing, Visualisation And Categorisation Of Network Traffic As Benign Or Dangerous.

Published:

24-8-2026

Issue:

Vol. 26 No. 8 (2026)

Page Nos:

1064 - 1070

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite