ANOMALY-BASED INTRUSION DETECTION SYSTEMS USING MACHINE LEARNING ALGORITHM

Aug 3, 2026

The increasing number of cyber-attacks increases the need for intrusion detection systems (IDS) capable of accurately and efficiently detecting and classifying attacks. Machine learning methods can learn from network data to discover patterns associated with intrusions or anomalies. This research aims to find a highly accurate, error-tolerant classifier for identifying anomalous traffic in a network intrusion detection dataset. Seven machine learning techniques are evaluated using accuracy, precision, recall, and F1-score, enabling comparison of strengths and weaknesses to support stronger, more dependable defenses for digital networks.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages90

Reading Time3 h 47 min

AccessFree to download

1. INTRODUCTION

1.1 BACKGROUND

1.2 INTRODUCTION TO INTRUSION

1.2.1 Intrusion Detection

1.2.2 Various Methods of Intrusion

1.3 PROCESS OF USING TOOLS & METHODS

1.3.1 Intrusion Detection System

1.3.2 Classification of Intrusion Detection System

1.4 THESIS OBJECTIVES

1.5 THESIS CONTRIBUTION

1.6 PROPOSED WORK

1.7 EXPECTED RESULTS

1.8 THESIS OUTLINE

REPUBLIC OF TÜRKİYE ALTINBAŞ UNIVERSITY Institute of Graduate Studies Electrical and Computer Engineering ANOMALY-BASED INTRUSION DETECTION SYSTEMS USING MACHINE LEARNING ALGORTIHIM Karrar Ali Awad AL-JUBOORI Master’s Thesis Supervisor Prof. Dr. Osman Nuri UÇAN Istanbul, 2023 ANOMALY-BASED INTRUSION DETECTION SYSTEMS USING MACHINE LEARNING ALGORITHM Karrar Ali Awad AL-JUBOORI Electrical and Computer Engineering Master’s Thesis ALTINBAŞ UNIVERSITY 2023 This thesis title” ANOMALY-BASED INTRUSION DETECTION SYSTEMS USING MACHINE LEARNING ALGORITHM” prepared by KARRAR ALI AWAD AL-JUBOORI and submitted on 5/4/2023 has been accepted unanimously for the degree of Master of Science in Electrical and Computer Engineering. Prof. Dr. Osman Nuri UÇAN Supervisor Thesis Defence Committee Members: Prof. Dr. Osman Nuri UÇAN Asst. Prof. Dr. Abdullahi Abdu IBRAHIM Department of Electrical and Electronic Engineering, Altinbas University Department of Computer __________________ Engineering, Altinbas University Asst. Prof. Dr. Serdar KARGIN Department of Electronics and Communication Engineering, Istanbul Arel University __________________ I hereby declare that this thesis meets all format and submission requirements ofa Master’s thesis. Submission date of the thesis to institute of Graduate Studies: / / By signing this declaration, I vouch for the fact that all information and resources used in this capstone project were acquired with the utmost adherence to moral standards and academic regulations. All references listed in the Reference List and vice versa have been cited in accordance with the aforementioned standards of conduct. I also attest to the accurate citation of all conclusions and borrowed ideas throughout the article. Karrar Ali Awad AL-JUBOORI Signature DEDICATION To My Country “IRAQ” To my dear father and my role model in life, who inspired me to continue my scientific path, and to my beloved wife and daughter, who accompanied me at all times and were the blessings of Support and help…My Mother, Brothers and Sisters. ABSTRACT ANOMALY-BASED INTRUSION DETECTION SYSTEMS USING MACHINE LEARNING ALGORITHM AL-JUBOORI, Karrar Ali Awad M.Sc., Electrical and Computer Engineering, Altinbas University, Supervisor: Prof. Dr. Osman Nuri UÇAN Date: April / 2023 Page: 85 The increasing number of cyber-attacks highlights the need for improved intrusion detection systems (IDS) that can detect and classify attacks accurately and efficiently. Machine learning (ML) techniques have been shown to be effective in this regard, as they can learn from data and identify patterns that may indicate an intrusion or anomaly. In this research, we want to find a highly accurate and error-tolerant classifier for identifying aberrant traffic in a network intrusion detection dataset. We will evaluate the effectiveness of seven different machine learning techniques on a given dataset by using various evaluation metrics such as accuracy, precision, recall, and F1-score. These measurements allow us to evaluate the effectiveness of various algorithms and offer insightful information about the advantages and disadvantages of each approach. Our ultimate goal is to select the best-performing classifier that can accurately detect anomalous traffic with minimal error. By identifying an effective classifier for NIDS, our goal is to aid in the creation of stronger, more dependable systems that can defend against cyberattacks and guarantee the safety of digital networks. Keywords: Instruction Detection System (IDS) , Network IDS, Machine Learning, Cyber Security. TABLE OF CONTENTS Pages [ABSTRACT ........................................................................................................................ vi](ABSTRACT ........................................................................................................................ vi) [LIST OF FIGU

FAQs

What problem does the thesis address in network security?

It addresses the need for improved intrusion detection systems that can detect and classify cyber-attacks accurately and efficiently, especially by identifying anomalous traffic.

How does the research use machine learning to detect intrusions?

It evaluates seven machine learning techniques that learn from network data and identify patterns indicating intrusions or anomalies.

What evaluation metrics are used to compare the classifiers?

The study compares approaches using accuracy, precision, recall, and F1-score to assess performance and error tolerance.

Aug 3, 2026

Upload by Evangeline