ITC-571 Annotated Bibliography: Hybrid SVM and ELM for IDS Analysis

This annotated bibliography focuses on a research paper discussing a multi-level hybrid Support Vector Machine (SVM) and Extreme Learning Machine (ELM) model for Intrusion Detection Systems (IDS), enhanced by a modified K-means algorithm. The hybrid solution aims to improve network security by effectively detecting vulnerabilities and intrusions. The model employs four SVMs to categorize network traffic into DoS, U2R, R2L, or Normal, while an ELM classifier identifies vulnerabilities within the Probe category. Pre-processing of the training data and the application of a modified K-means algorithm on five datasets (Normal, DoS, Probe, R2L, and U2R) enhance the performance of both SVM and ELM. The hybrid model demonstrates high accuracy, achieving a detection rate of up to 95.75%, particularly effective for Normal and DoS traffic, with improved detection rates for U2R and R2L attacks. Desklib offers a platform for students to access similar solved assignments and past papers.

Multi-level hybrid SVM (Support Vector
Machine) and ELM (Extreme Learning
Machine) based on modified K-means for
IDS
The increase in connectivity between computers increases the need of advancement in network
security in order to detect vulnerabilities and intrusion for enhancing cyber security. The real
intrusion detection problems need to effective analysed and detected with the use of machine
learning techniques. The implication of machine learning helps in improving efficiency in
detecting unknown and known attacks.
SVM classifier is based on statistical learning theories. ELM transforms the learning problem
into a simple linear system. This hybrid solution facilitates effective intrusion detection that
effectively analyse traffic data network. The denial of service and Probe are static at the initial
stage of the model. As compared to the normal traffic, the number of attacks on the network is a
little fraction. This hybrid solution utilised four SVM’s in order to divide instances as DoS, U2R,
R2L or Normal. The ELM classifier is used with intention to detect vulnerabilities under the
Probe category.
The pre-processing of training data set takes place. Then the solution builds five sets named as
Normal, DoS, Probe, R2L and U2R. Further, a modified K-means algorithm is applied on every
dataset in order make enhancement in existing 5 sets. The implementation of K-means algorithm
supports in enhancing the performance of SVM and ELM. This solution has the effective
detection rate for Normal and DoS while the better detection rates for U2R and R2L. This hybrid
model achieves a top performance having high accuracy rate of up to 95.75%.

Al-Yaseen, W. L., Othman, Z. A., & Nazri, M. Z. A. (2017). Multi-level hybrid support vector
machine and extreme learning machine based on modified K-means for intrusion detection
system. Expert Systems with Applications, 67, 296-303.