CNN Implementation for IDS

Real world data is more complex in nature and requires proper handling to use with machine learning techniques. Handling large volumes of data is the key aspect in deep learning (DL). As new attacks are evolving, traditional Intrusion detection systems (IDS) are unable to handle those complex attack...

Full description

Saved in:
Bibliographic Details
Published in2021 3rd International Conference on Advances in Computing, Communication Control and Networking (ICAC3N) pp. 970 - 975
Main Authors Varanasi, Venkata Ramani, Razia, Shaik
Format Conference Proceeding
LanguageEnglish
Published IEEE 17.12.2021
Subjects
Computational modeling
Convolutional neural network
Deep learning
Deep neural network
Intrusion detection
Neural networks
Object detection
Training
Transfer learning
Online AccessGet full text

Cover

More Information
Summary:Real world data is more complex in nature and requires proper handling to use with machine learning techniques. Handling large volumes of data is the key aspect in deep learning (DL). As new attacks are evolving, traditional Intrusion detection systems (IDS) are unable to handle those complex attacks. With successful applications of convolutional neural networks (CNN) in other fields like object detection, face recognition, healthcare, this paper proposes CNN based IDS using CIC-IDS-2017 dataset. The performance of the model is compared with that of a deep neural network (DNN). The numerical features are converted into gray scale images and fed to the model. The model is repeated with different values of hyper parameters and results are presented. CNN model can be considered as useful for transfer learning as it reduces training time and gives better accuracy by considering the correlation between the features. The authors implemented CNN model with the objective of applying it to homogeneous and heterogeneous transfer learning in the future. CNN model attained an accuracy of 99% as compared to DNN with 98%.
DOI:10.1109/ICAC3N53548.2021.9725426