![Related Work
• The authors in [1] proposed an ensemble model comprising Gradient Boosting and Random Forest as base
learners and Naive Bayes as a meta learner for detecting zero-day vulnerabilities in network traffic.
• Similarly, a deep stacked autoencoder followed by a Long-Short-Term Memory model is used in [2] to
significantly enhance ransomware stratification accuracy.
8](https://image.slidesharecdn.com/anomalydetectioninnetworktrafficusingmachinelearning-250129152311-24e36a92/85/Anomaly-Detection-in-Network-Traffic-using-Machine-Learning-pptx-8-320.jpg)
![15
References
[1] Nkongolo, M. N. W. (2023). Zero-day vulnerability prevention with recursive feature elimination and
ensemble learning. Cryptology ePrint Archive.
[2] Nkongolo, M., & Tokmak, M. (2024). Ransomware detection using stacked autoencoder for feature
selection. arXiv preprint arXiv:2402.11342.](https://image.slidesharecdn.com/anomalydetectioninnetworktrafficusingmachinelearning-250129152311-24e36a92/85/Anomaly-Detection-in-Network-Traffic-using-Machine-Learning-pptx-15-320.jpg)
Anomaly Detection in Network Traffic using Machine Learning.pptx
- 1. DEPARTMENT OF COMPUTER SCIENCE AND INFORMATION TECHNOLOGY, WOMEN UNIVERSITY OF AZAD JAMMU & KASHMIR, BAGH. Anomaly Detection in Network Traffic using Machine Learning Presented By: Maria Khalil Soomie Aftab Supervised By: Dr. Jawad-ur-Rehman Chughtai
- 2. Introduction 1 Motivation & Scope 2 Goals & Objectives 4 Related work 3 Agenda Tools & Techniques 6 Gantt Chart 5
- 3. Introduction
- 4. 4 • The ever-growing threat landscape and the increasing volume of network traffic make anomaly detection a crucial aspect of cybersecurity. • Machine learning (ML) offers powerful tools to identify unusual patterns that might indicate malicious activity. • The focus of this project will be to develop a machine leanring-based solution for anomaly detection in network traffic. Introduction
- 6. Motivation & Scope • In recent years, cyber attacks have emerged as one of the most pervasive and damaging cybersecurity threats facing organizations worldwide. • These malicious attacks encrypt valuable data and demand hefty ransom payments in exchange for decryption keys, causing significant financial losses, operational disruptions, and reputational damage to affected entities. • Traditional security measures, such as firewalls and antivirus software, often struggle to detect and mitigate cyber attacks effectively due to their rapidly evolving nature and sophisticated evasion techniques. • There is a need to develop a machine leanring-based solution for anomaly detection in network traffic 6
- 7. Related Work
- 8. Related Work • The authors in [1] proposed an ensemble model comprising Gradient Boosting and Random Forest as base learners and Naive Bayes as a meta learner for detecting zero-day vulnerabilities in network traffic. • Similarly, a deep stacked autoencoder followed by a Long-Short-Term Memory model is used in [2] to significantly enhance ransomware stratification accuracy. 8
- 10. 10 Goals • Develop a robust machine learning-based anomaly detection system capable of accurately identifying and classifying cyber attacks within network traffic. • Enhance the proactive defense capabilities of organizations against cyber attacks. • Improve the accuracy and efficiency of anomaly detection by leveraging machine learning algorithms to analyze network traffic patterns and identify anomalous behavior indicative of cyber activity. • Collect and preprocess a comprehensive dataset of labeled network traffic samples. • Transform categorical features using appropriate feature encoding technique and compute feature importance and select the most important features using feature selection techniques. • Normalize the final feature set using appropriate feature scaling technique and develop and train different machine learning models using the final feature set. • Optimize their performance metrics such as accuracy, precision, recall, and F1 score and select the best model for final predictions. Objectives
- 11. Gantt Chart
- 12. 12 Gantt Chart
- 14. 14 Tools & Technologies • For documentation, we will use Overleaf template. • For coding, we will use python. • For deployment/webpage, we will use Streamlit.
- 15. 15 References [1] Nkongolo, M. N. W. (2023). Zero-day vulnerability prevention with recursive feature elimination and ensemble learning. Cryptology ePrint Archive. [2] Nkongolo, M., & Tokmak, M. (2024). Ransomware detection using stacked autoencoder for feature selection. arXiv preprint arXiv:2402.11342.
- 16. Thanks