Browsing by Author "Surasit Songma"

Now showing 1 - 3 of 3
  • Default Image
    Item
    Classification via k-means clustering and distance-based outlier detection
    (2012) Surasit Songma; Witcha Chimphlee; Kiattisak Maichalernnukul; Parinya Sanguansat
    We propose a two-phase classification method. Specifically, in the first phase, a set of patterns (data) are clustered by the k-means algorithm. In the second phase, outliers are constructed by a distance-based technique and a class label is assigned to each pattern. The Knowledge Discovery Databases (KDD) Cup 1999 data set, which has been utilized extensively for development of intrusion detection systems, is used in our experiment. The results show that the proposed method is effective in intrusion detection. © 2012 IEEE.
  • Default Image
    Item
    EXTENDING NETWORK INTRUSION DETECTION WITH ENHANCED PARTICLE SWARM OPTIMIZATION TECHNIQUES
    (Academy and Industry Research Collaboration Center (AIRCC), 2024) Surasit Songma; Watcharakorn Netharn; Siriluck Lorpunmanee
    The present research investigates how to improve Network Intrusion Detection Systems (NIDS) by combining Machine Learning (ML) and Deep Learning (DL) techniques, addressing the growing challenge of cybersecurity threats. A thorough process for data preparation, comprising activities like cleaning, normalization, and segmentation into training and testing sets, lays the framework for model training and evaluation. The study uses the CSE-CIC-IDS 2018 and LITNET-2020 datasets to compare ML methods (Decision Trees, Random Forest, XGBoost) and DL models (CNNs, RNNs, DNNs, MLP) against key performance metrics (Accuracy, Precision, Recall, and F1-Score). The Decision Tree model performed better across all measures after being fine-tuned with Enhanced Particle Swarm Optimization (EPSO), demonstrating the model's ability to detect network breaches effectively. The findings highlight EPSO's importance in improving ML classifiers for cybersecurity, proposing a strong framework for NIDS with high precision and dependability. This extensive analysis not only contributes to the cybersecurity arena by providing a road to robust intrusion detection solutions, but it also proposes future approaches for improving ML models to combat the changing landscape of network threats. © (2024), (Academy and Industry Research Collaboration Center (AIRCC)). All Rights Reserved.
  • Default Image
    Item
    Optimizing Intrusion Detection Systems in Three Phases on the CSE-CIC-IDS-2018 Dataset
    (Multidisciplinary Digital Publishing Institute (MDPI), 2023) Surasit Songma; Theera Sathuphan; Thanakorn Pamutha; S. Songma; Department of Information Technology, Faculty of Science and Technology, Suan Dusit University, Bangkok, 10300, Thailand; email: surasit_son@dusit.ac.th
    This article examines intrusion detection systems in depth using the CSE-CIC-IDS-2018 dataset. The investigation is divided into three stages: to begin, data cleaning, exploratory data analysis, and data normalization procedures (min-max and Z-score) are used to prepare data for use with various classifiers; second, in order to improve processing speed and reduce model complexity, a combination of principal component analysis (PCA) and random forest (RF) is used to reduce non-significant features by comparing them to the full dataset; finally, machine learning methods (XGBoost, CART, DT, KNN, MLP, RF, LR, and Bayes) are applied to specific features and preprocessing procedures, with the XGBoost, DT, and RF models outperforming the others in terms of both ROC values and CPU runtime. The evaluation concludes with the discovery of an optimal set, which includes PCA and RF feature selection. © 2023 by the authors.