Abstract:
The Internet of Things (IoT) combines sensors and other small devices interconnected locally and via the Internet. Specifically, IoT devices collect data from the environment through sensors, analyze it, and respond to the actual through controllers. The integration of these devices can be seen in various areas like home appliances, healthcare, control systems, etc. On the other hand, massive digital data can drive system performance, and data security is also a serious concern. Therefore, anomaly detection (AD) is necessary to prevent network security infractions and system attacks. Several Artificial Intelligence (AI)-based anomaly detection methods have been designed with higher detection performance; however, they are still “complex” models that are hard to interpret and explain. This chapter proposes a hybrid learning model for AD in IoT with Explainable Artificial Intelligence to enhance the perspective and explainable results. The proposal’s application uses a well-known traffic traces dataset (https://www.kaggle.com/datasets/francoisxa/ds2ostraffictraces). Our code and dataset are added to https://github.com/himanshubeniwal/ml-xai.