Intrusion Detection in
Networks
Using Machine Learning
Techniques
Supervised By
Dr. Anisur Rahman
Associate Professor
Department of Computer Science &
Engineering
East West University, Dhaka, Bangladesh
Ali Al Absar Student ID: 2018-2-5 0 - 006
Alvee
Student ID: 2019-1-5 0 - 027
Shahad Israq
Asif Iqbal Student ID: 2019-1-5 0 - 0 5 4
Anik Roy
Student ID: 2019-2-5 0 - 026
Introduction
Overview of ML & IDS
Supervised Learning
Unsupervised Learning
Using Algorithms for Intrusion Detection
Overview of the Process
Flowchart
Data Description
Confusion Matrix
Cross Validation Score
Findings & Result
Limitations & Challenges
Future Direction
Conclusion
Introductio
n
Machine learning transforms
Network Intrusion Detection
Systems, empowering them to
adapt and detect evolving cyber
threats with greater accuracy and
intelligence.
Problem Definition
Objectives
Scope
Thesis Structure
Machine Learning
Supervised Learning
Unsupervised Learning
Reinforcement Learning
Intrusion Detection
Systems (IDS)
Signature-Based Detection
Anomaly-Based Detection
Supervised Learning
Linear Regression
Logistic Regression
Decision Trees
Random Forest
Support Vector Machines (SVM)
K-Nearest Neighbors (KNN)
Unsupervised
Learning
K-Means Clustering
Hierarchical Clustering
Principal Component Analysis (PCA)
Principal Component Analysis (PCA)
t-Distributed Stochastic Neighbor
Embedding (t-SNE)
Algorithms Used for
Intrusion Detection
Decision Trees
Logistic
Regression
K-Nearest Neighbors
(KNN)
Overview of the
Process
D a t a C o l l e c t i o n
Data Preprocessing
Feature Selection
Data Normalization
Model Training
Hyperparameter Tuning
Model Evaluation
Flowchart of
Methodology
Dataset Description
Source of the Dataset
Features in the Dataset
Class Distribution
Data Preprocessing
Confusion Matrix
Cross-Validation Scores
Findings & Result
• Decision Trees
• Logistic Regression
• K-Nearest Neighbors
• Feature selection with RFE
• Hyperparameter tuning
Limitation and Challenges
Computational Complexity
Class Imbalance
Dataset Limitation
Future
Directions
• R e a l -T i m e
Implementation
• Class Imbalance Handling
• Deep Learning Approaches
• Ensemble Methods
Conclusion
This thesis presented a comprehensive study of machine learning
models for network intrusion detection. Several models, including
L o g i s t i c R e g r e s s i o n , K- N e a r e s t N e i g h b o r s , D e c i s i o n T r e e s w e r e
trained and evaluated on a publicly available dataset. The study
focused on model performance, feature selection, hyper
parameter tuning with Optuna for optimization, and dealing with
class imbalance.
Thank
