AI Based Anomaly Detection and Threat Prevention

Managing and monitoring the functioning of distributed systems is a vital activity in today's age. With hundreds of thousands of things to monitor, anomaly detection can assist in identifying where an error is occurring, improving root cause investigation, and recognizing the potential external and

Project Title

AI Based Anomaly Detection and Threat Prevention

Project Area of Specialization

Artificial Intelligence

Project Summary

Managing and monitoring the functioning of distributed systems is a vital activity in today's age. With hundreds of thousands of things to monitor, anomaly detection can assist in identifying where an error is occurring, improving root cause investigation, and recognizing the potential external and internal threats before they turn into attacks. Anomaly Detection is a method for identification and diagnosis of critical incidents, such as a technological problem, or prospective opportunities, such as a shift in system's behavior, as well as alerting the concerned authority to act. Learning-based systems for identifying cyber threats have been facilitated by the development of artificial intelligence (AI) techniques and algorithms, and they have shown substantial outcomes in several studies. However, due to the ever-changing nature of cyber intrusions, protecting IT systems from threats and fraudulent activity in networks remains a major challenge. The impact on security professionals will be determined whether these advances lead researchers into understanding or addressing problems with network defense practices at scale. The aim is to provide improved artificial intelligence-based cybersecurity which includes traffic analysis on the heterogeneous dataset. Furthermore, any odd actions will be detected. We have devised a solution for threat detection using artificial intelligence behavior analysis-based anomaly recognition.

Project Objectives

This project will meet the following objectives:

• In threat analysis, IPS/IDS are still utilized, but they are signature-based, whereas we will use Artificial Intelligence Behavior Analysis for detection.

• By completing the proposed FYP, we will be able to see how the amount and variety of network traffic affect the accuracy of intrusion detection

• Recognizing the difference between true positive and false positive alerts, allowing security analysts to respond to real cyber threats more effectively

Project Implementation Method

The system's initial preprocessing phase seeks to convert raw data into concise inputs to be fed into the NEURAL NETWORK algorithm. The preprocessed data is supplied into the artificial neural networks for data learning, and the ANN performs learning to identify the best accurate model. Finally, in threat detection, the selected ANN model uses the trained model to mechanically classify each security raw event, and the dashboard displays the results in an easily interpretable manner. Only real warnings will be identified by security analysts to reduce fake ones. Through the integration of Artificial Intelligence, we will be able to improve existing solutions by employing self-learning capabilities and implementing feature engineering. This strategy is yet to be implemented in Pakistan as our government lacks a deep understanding of risk perception or threats at scale. For data learning, the preprocessed data is fed into artificial neural networks, and each ANN learns to find the most correct model. Finally, each ANN model utilizes the trained model to mechanically categorize each security raw behavior for threat detection, and the dashboard displays the entire log of threat detection. Additionally, it will also display the respective IP addresses of the attacker/threat source and victim device.

Benefits of the Project

• The traditional signature-based anomaly detection approaches due to the emergence of polymorphic attacks and their limitations are not reliable for present and upcoming cybersecurity challenges. To overcome the cybersecurity gap, we propose a solution based on Artificial Intelligence and behavior analysis anomaly detection.

• Machine learning-based methods for detecting unusual patterns can help detect emerging cyber threats.

• This methodology can allow superior categorization for true alerts when compared to traditional machine learning methods, it can significantly reduce the amount of false-positive alerts that analysts get

Technical Details of Final Deliverable

• Project definition • Reference Research Papers • Detailed Proposal • Feature List • Initial Proposal • System diagram • Datasets Analysis • Attack and Normal Packet analysis • Normalized and ready to use data • EDA Completed• Working on ML algorithm and program. • AI pipeline program and analytics dashboard ready • Trained and tested NEURAL NETWORK algorithm • Analytics dashboard ready through Kibana• Working Project in executable and modified form • Final FYP Report • Multiple copies of FYP Report

Final Deliverable of the Project

Software System

Core Industry

IT

Other Industries

Security

Core Technology

Artificial Intelligence(AI)

Other Technologies

Others

Sustainable Development Goals

Partnerships to achieve the Goal

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com