AI-Driven Anomaly Detection Technologies to Support Emerging Smart Applications
1Nanjing Forestry University, Nanjing, China
2Abdul Wali Khan University, Mardan, Pakistan
3Macquarie University, Sydney, Australia
AI-Driven Anomaly Detection Technologies to Support Emerging Smart Applications
Description
Recently, many emerging smart applications have been proposed and applied in diverse platforms, such as Internet of Things (IoT), Internet of Medical Things (IoMT), Web of Things (WoT), and other cyber-physical devices. As a widely used AI-driven technology, supervised learning has played a huge role in time-series signal processing, text analysis, and biometric recognition, which is used to infer unseen instances via training a model from historical data.
However, the performances of supervised learning are often limited by the disturbance of abnormal data. It is difficult to collect all classes from limited historical instances. The future instances may be from an unknown class that does not exist in the historical data. Generally, these instances deviate from the main distribution of the known classes and are called anomalies or novelties. The anomalies or novelties widely exist in many real applications, such as illegal intrusion in internet of services, irregular visiting in edge computing, and abnormal events in IoT, to name just a few. In addition, the historical instances from massive IoT devices and applications would be corrupted by minor noises. It is impractical to collect thousands of uniformly distributed data from heterogeneous distributed devices and associated smart applications. It would be significantly affected by the noises to learn a supervised model by using a relatively small dataset. From the above two aspects, anomaly detection is critical for AI-driven technologies to support IoT and the associated smart applications. As a challenge in the community of machine learning, anomaly detection has been researched for several decades.
This Special Issue will focus on using state-of-the-art anomaly detection methods to support the emerging technologies for IoT and smart applications, e.g. Smart Health, Smart Mobility & Connected Vehicles, Smart City, Smart Farming, Smart Transportation, Smart Agriculture, Smart Building/Infra Structure, Smart Factories, Smart Energy, and Human–Machine Dialogue. This Special Issue aims to encourage further research and development of anomaly detection for supporting emerging technologies and to provide a unique opportunity to allow researchers in several domains of computing area to contribute to anomaly detection and its applications in IoT and smart applications. Original research and review articles are welcome.
Potential topics include but are not limited to the following:
- Architectures and platforms for IoT and IoMT with the capabilities of anomaly processing and data fusion
- Edge and mist-based platforms with anomaly information mining
- Real-time algorithms and methods for Internet of Things (IoT) and Internet of Medical Things (IoMT)
- Methodologies for anomaly detection-based smart health and fitness
- Efficient anomaly detection for synchronous multichannel signal processing in the Internet of Things (IoT) architectures or systems
- Emerging techniques for smart neural dialogue generation based on the anomaly technologies
- Bio-signal real-time processing in IoMT
- Anomaly information mining for IoT and IoMT applications
- AI-driven and smart algorithm-enabled edge computing for IoT and IoMT
- AI-driven data interpretation in the scenarios of IoT and IoMT