Lmproved classification of adversarial network intrusion using ensemble clustering based data transformation

Name: Pongsakorn Tatongjai

keyword: Intrusion Detection

MeSH: Computer networks -- 288056

Classification :.DDC: e-Theses

; Adversarial Attack, Machine Learning, Feature Transformation, Ensemble Clustering

MeSH: Computer networks -- Management -- 288057

MeSH: Machine learning -- 288058

MeSH: Cluster analysis -- 12879

LCSH: Computer networks -- 288056

LCSH: Computer networks -- Management -- 288057

LCSH: Machine learning -- 288058

LCSH: Cluster analysis -- 12879

Abstract: A surge in usage of communication networks for business and information services has increased the risk of attacks to both individuals and organizations. In response, a network-based intrusion detection system (NIDS) has been developed as one of the solutions in an effort to defend against such attacks and has proven to be effective against known intrusions. Data scientists and security practitioners are working to improve machine-learning based NIDS to become more potent for adversarial attacks that have been obfuscated to mask patterns of intrusive traffics. This research spotlights non-payload connections at the TCP level which are applied to different network applications. Three new filter models, ECT Subspace, ECTNoise and ECT have been proposed contrary to the wrapper method in the benchmark dataset. The new models transform the feature space without knowledge of class labels and are developed using ensemble clustering and three different ensemble generation strategies. According to the empirical study of published datasets and four classification algorithms, new models outperform the original wrapper and other filter models. The first experiment summarizes this by using basic classification on legitimate and direct attacks. The second targets the robustness to recognize obfuscated intrusions. As a guideline, analysis of algorithmic parameters, i.e., ensemble size and level of noise has been provided for practical use.

Mae Fah Luang University. Learning Resources and Educational Media Center

Address: CHIANG RAI

Email: library@mfu.ac.th

Name: Natthakan Iam-on

Role: Advisor

Created: 2021

Modified: 2022-07-05

Issued: 2022-07-05

วิทยานิพนธ์/Thesis

application/pdf

CallNumber: e-Theses

eng

DegreeName: Master of Science

Level: Master's Degree

Descipline: Information Technology

Grantor: Mae Fah Luang University

©copyrights Mae Fah Luang University

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	01 b.134871 6151301004 (4-7-65).pdf	2.85 MB	16	2026-05-26 18:16:01

ใช้เวลา

0.034059 วินาที