The Spectrum of AI Applications

We are living in exciting and amazing times in which AI is reshaping the future and transforming our lives – in many good ways. What has made these AI advances possible is not only the AI Innovative technologies that have matured but the convergence of AI with multiple innovative technologies that unleash new, unprecedented waves of transformation. These technologies include computational power, 5G, Internet of Things, augmented reality, virtual reality and robotics.

One area of AI that started decades ago, and has matured and become a primary form of machine learning is neural networks. These have evolved into various types of neural nets including deep learning and are continuing to improve. Neural network architectures are also achieving more capabilities in respect to accessing deeper insights in data, detecting anomalies, and providing predictive analytics. As a whole, the neural networks of today are more capable than ever before.

This presentation will focus on how AI, in particular neural networks, are being used in signal processing. However, AI and neural networks are not a solution to every problem. Therefore, we will discuss where neural networks are effective and have shown superior performance, compared to traditional approaches, as well as where they have underperformed. We will present appropriate uses and include some examples. The presentation will start with a brief overview of neural network techniques, and then introduce the applications to signal processing of neural networks, which include spectrum sensing, anomaly detection and dynamic spectrum access, and end with some future directions such as spectrum management.