Home // ICSEA 2019, The Fourteenth International Conference on Software Engineering Advances // View article

Denoising Autoencoder with Dropout based Network Anomaly Detection

Authors:
Safa Mohamed
Ridha Ejbali
Mourad Zaied

Keywords: Anomaly detection; NIDS; Denoising Autoencoder; NSL-KDD

Abstract:
A Network Intrusion Detection System (NIDS) plays an important role in ensuring information security. It helps system administrators identify and detect malicious activities in their companies. Many techniques have been devised by researchers to achieve reliable detection of anomalies. It is thus a challenging task to determine a network anomaly more accurately. To solve this problem, we propose a Denoising-Autoencoder (DAE) with a Dropout based network anomaly detection method because it forces the extraction of intrinsic features so as to increase the detection accuracy. A popular NSL-KDD dataset is used for the training and evaluation of our approach. The performance of our approach takes into consideration different metrics such accuracy, precision, recall, f-measure values and the detection rate. Experimental results show that our approach performs better than other detection methods, especially when we use a single hidden layer with 8 neurons.

Pages: 98 to 103

Copyright: Copyright (c) IARIA, 2019

Publication date: November 24, 2019

Published in: conference

ISSN: 2308-4235

ISBN: 978-1-61208-752-8

Location: Valencia, Spain

Dates: from November 24, 2019 to November 28, 2019