Securing Industrial IoT Networks Using Conv2D-Attention Approach with Softmax Classifier
Bhavya Bhola, Ayan Raza, Rajeev Kumar · 2024
In this paper, we propose Conv2D-attention based intrusion detection systems (IDS) to secure the network in the industrial internet-of-things (IIoT). For this purpose, convolution on image style input data is applied in the feature extraction section to extract the spatial features. Further, multi-head attention is employed to attend different aspects of input data simultaneously. Furthermore, the feature extraction section is connected to resnet style deep neural networks to provide robust classification. In addition, the softmax classifier is used to predict the label. Our proposed Conv2D attention system performs better as compared to existing system models available in literature.