Dissertation/Thesis Abstract

Benchmarks for Evaluating Anomaly-Based Intrusion Detection Solutions
by Miller, Nicholas J., M.S., California State University, Long Beach, 2018, 30; 10752128
Abstract (Summary)

Anomaly-based Intrusion Detection Systems are critical components of modern security systems. They often rely on Machine Learning (ML) to detect potential attacks and have gained increased popularity over time, due to new technologies and dangers. There are many proposed anomaly-based systems using different ML algorithms and techniques, however there is no standard benchmark to compare these based on quantifiable measures.

We have proposed a benchmark that measures both accuracy and performance to produce objective metrics that can be used in the evaluation of each algorithm implementation. In this paper, the benchmark will be used to compare four different ML algorithms (Naive Bayes, Support Vector Machines, Neural Networks, and K-means Clustering) on the NSL-KDD dataset. The experimental results show the differences in accuracy and performance between these algorithms on the dataset, and also how this benchmark can be used to create useful metrics for comparisons.

Indexing (document details)
Advisor: Aliasgari, Mehrdad
Commitee: Ebert, Todd, Zhang, Wenlu
School: California State University, Long Beach
Department: Computer Engineering and Computer Science
School Location: United States -- California
Source: MAI 57/05M(E), Masters Abstracts International
Source Type: DISSERTATION
Subjects: Computer science
Keywords: Anomaly-based, Benchmark, Cybersecurity, Intrusion detection systems, Machine learning, Nsl-kdd
Publication Number: 10752128
ISBN: 9780355921311
Copyright © 2019 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy
ProQuest