Anomaly-Based Network Intrusion Detection Using Hybrid CNN, Bi-LSTM Deep Learning Techniques

The network anomaly and threat detection are essential components of the cyber security field due to continually growing network traffic and the frequent emergence of new attack types. Deep learning (DL) has become increasingly important in anomaly detection in recent years, particularly in the fiel...

Full description

Saved in:
Bibliographic Details
Published in:2024 4th International Conference on Innovative Research in Applied Science, Engineering and Technology (IRASET) pp. 1 - 6
Main Authors: Akkepalli, Srinivas, Sagar, K
Format: Conference Proceeding
Language:English
Published: IEEE 16.05.2024
Subjects:
Bi-LSTM
CNN
Computational modeling
Computer crime
Deep learning
NSL KDD
Recurrent neural networks
RNN
Statistical analysis
Telecommunication traffic
Threat assessment
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The network anomaly and threat detection are essential components of the cyber security field due to continually growing network traffic and the frequent emergence of new attack types. Deep learning (DL) has become increasingly important in anomaly detection in recent years, particularly in the field of cyber security. While machine learning (ML) algorithms and conventional known rules-based or signaturebased methods for detecting anomalies have been employed, these approaches are only effective in detecting point anomalies, cannot recognize for adjust to the evolving novel patterns in the data. As a result, in recent years, cyber security has taken center stage. Identification of potential attack patterns requires careful observation and analysis of network traffic data. This research includes a variety of methodologies from the fields of computers, statistics, information, and technology, including machine learning. This paper describes a DL model more precisely, a Bidirectional LSTM that combines the unique benefits of RNN (Recurrent Neural Network) and a convolution neural network. The suggested model has a relatively low FPR (False Positive Rate), a high detection rate and high accuracy. Ultimately, this study assesses the efficacy of deep learning approaches and the suggested model in the domain of network anomaly detection by comparing them to the most recent machine learning models.
DOI:10.1109/IRASET60544.2024.10548678