Intrusion Detection System Using Machine Learning Thesis Ideas
Entire thesis ideas and topics will be written right from the introduction to conclusion as per scholars’ request by our experts. Thesis topics will be suggested based on your unique research goal. From scratch we prepare your research work. We guarantee that your thesis work will be kept 100% confidential. The topics that we have worked are listed below.
- 1. An Efficient Network Intrusion Detection System for Distributed Networks using Machine Learning Technique
Keywords:
Network Intrusion Detection System (NIDS), Distributed Denial of Service (DDoS), Random Forest (RF), Machine Learning, distributed Networks, Accuracy, Support Vector Machine (SVM)
We propose a Big Data-based Distributed Denial of Service Network Intrusion Detection System and in our study the micro-batch data processing is engaged for traffic feature collection in the network collection module and Random Forest based classification method is utilized in traffic detection module for feature selection. To store a large amount of wary attacks we used Hadoop File System (HDFS). To suggest a solution we used S park. Our work can be compared with the ML methods like DT, PCARF, NB, SVM and LR to get a high accuracy.
- A Machine Learning-Based Intrusion Detection System for IoT Electric Vehicle Charging Stations (EVCSs)
Keywords:
Anomaly detection; cyber security; feature selection; Internet of Things (IoT); intrusion detection system (IDS); security
Our paper proposes a classifier method to detect malicious traffic in IoT environment by utilizing a ML method. We used a real IoT dataset that can obtained from real IoT traffic. We have been evaluated the multiple classifying methods and by utilizing our proposed method in IoT-based IDS engine that serves electric vehicle charging station that will bring stability and to remove a large number of cyberattacks.
- Adversarial Machine Learning for Network Intrusion Detection Systems: A Comprehensive Survey
Keywords:
Adversarial attacks, deep learning
We review about the recent literature on NIDS, adversarial attacks and network defences to observe the difference in adversarial learning against deep neural network in CV and NIDS. We offer a reader with a better understanding of DL based NIDS, adversarial attacks and defences. At first, we offer a classification of DL-based NIDS and converse the impact of taxonomy on adversarial attacks. Next, we have to evaluate the white-box and black-box adversarial attack on DNNs finally we defence against adversarial features.
- Machine-Learning-Based UAV-Assisted Agricultural Information Security Architecture and Intrusion Detection
Keywords:
Agricultural information security, convolutional neural network (CNN), geographic position information (GPI), intrusion detection, unmanned aerial vehicles (UAVs)
We aim to ensure the safe operation of agricultural information systems and to assure the data security of intelligent architecture. First the UAV-aided information acquisition system has studied also a double deep $Q$ -network (DDQN) for location deployment based on geography position information (GPI) to fastly optimize the location of UAV. Also CNN and LSTM were combined as CNN-LSTM method to construct IDS for AIoT for agriculture. Our proposed CNN+LSTM gives the best accuracy.
- A dependable hybrid machine learning model for network intrusion detection
Keywords:
XGBoost, Feature Importance, Dependability
We propose a new hybrid method that integrated the ML and DL methods to improve the detection rate while securing dependability. Our proposed method confirms efficient preprocessing by integrating SMOTE for data balancing and XGBoost for feature selection. We compare our methods to different ML and DL method to identify the best method to implement in pipeline.
- Implementation of Intrusion Detection System Using Various Machine Learning Approaches with Ensemble learning
Keywords:
Cyber-Attacks, Hyper parameters, Meta-Heuristic
We propose a novel approach to enhance the performance IDS in NSL-KDD dataset by utilizing the multiple meta-heuristic and ML methods. The multiple meta-heuristic methods can be used to optimize the hyper-parameter of ML methods like RF, CART, SVM and MLP. Our paper also notes the important capability of metaheuristic methods to optimize the IDS models and the effectiveness of ML based solutions.
- Adversarial Attack of ML-based Intrusion Detection System on In-vehicle System using GAN
Keywords:
GAN, In-vehicle networks
Our study proposes a Generative Adversarial Network (GAN) based method to create adversarial attacks that accomplished by passing ML-based IDS in vehicle network. We contain preprocessing an automotive hacking dataset to train a GAN-based method and evaluate thee generated attack by utilizing accuracy metrics. In addition the t-SNE visualization reveals the effective new adversarial attacks to fortify the security.
- Intrusion Detection in IoT leveraged by Multi-Access Edge Computing using Machine Learning
Keywords:
Multi-Access Edge Computing, Network security
We concentrate by utilizing ML and DL methods along with feature selection method to detect cyberattack effectively at the edge of IoT by manipulating multi-access edge computing. We use ANOVA and embedded feature selection methods and use different ML methods like DT, RF, LightGBM, ANN, KNN and XGB on UNSW-NB15 dataset. Our classifier can give the best accuracy and LightGBM is the accurate among all.
- A machine-learning-based Intrusion detection for IIoT infrastructure
Keywords:
Industrial Internet of Things, Naive Bayes
Our paper proposes an intrusion detection system for IIoT based on ML methods. To categorize the incoming traffic as normal or malicious by utilizing the methods like DT, RF and NB methods. Our dataset was preprocessed and the features were retrieved by train and test the suggested IDS. The decision tree method can achieve the best accuracy rate. Our proposed IDS are estimated to improve the security of IIoT system and moderate the hazard of cyberattack.