Anomaly-Based DNN Model for Intrusion Detection in IoT and Model Explanation

Explainable Artificial Intelligence

Conference Paper (2023)
Author(s)

Bhawana Sharma (Manipal University Jaipur)

Lokesh Sharma (Manipal University Jaipur)

Chhagan Lal (TU Delft - Cyber Security)

Research Group
Cyber Security
Copyright
© 2023 Bhawana Sharma, Lokesh Sharma, C. Lal
DOI related publication
https://doi.org/10.1007/978-981-19-6661-3_28
More Info
expand_more
Publication Year
2023
Language
English
Copyright
© 2023 Bhawana Sharma, Lokesh Sharma, C. Lal
Research Group
Cyber Security
Bibliographical Note
Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.@en
Pages (from-to)
315-324
ISBN (print)
978-981-19-6660-6
ISBN (electronic)
978-981-19-6661-3
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

IoT has gained immense popularity recently with advancements in technologies and big data. IoT network is dynamically increasing with the addition of devices, and the big data is generated within the network, making the network vulnerable to attacks. Thus, network security is essential, and an intrusion detection system is needed. In this paper, we proposed a deep learning-based model for detecting intrusions or attacks in IoT networks. We constructed a DNN model, applied a filter method for feature reduction, and tuned the model with different parameters. We also compared the performance of DNN with other machine learning techniques in terms of accuracy, and the proposed DNN model with weight decay of 0.0001 and dropout rate of 0.01 achieved an accuracy of 0.993, and the reduced loss on the NSL-KDD dataset having five classes. DL models are a black box and hard to understand, so we explained the model predictions using LIME.

Files

978_981_19_6661_3_28.pdf
(pdf | 0.53 Mb)
- Embargo expired in 28-07-2023
License info not available