Objectives: Caching and multi-agent have found extensive use in practical intelligent Internet of Things (IoT) networks, thanks to advancements in wireless communications and artificial intelligence. These networks include mobile edge computing, smart cities, smart homes, smart grids, and industrial IoT. A centralised node is typically utilised to gather the status and reward of all the agents in these applications, and numerous agents interact with each other by transferring the training model or gradients via caching. On the other hand, this poses serious challenges to communication and raises the possibility of data leakage and islands. One possible solution is to use distributed optimization in cache-enabled multi-agent learning; another is federated learning, which is a common kind of distributed optimization that can overcome the data island problem. For intelligent IoT networks that make use of caches, studies on sophisticated distributed optimization have recently exploded in popularity.
Impact: When it comes to cache-enabled intelligent IoT networks, advanced distributed optimization using multi-agent learning still faces a number of basic obstacles. The application of other intelligent learning algorithms, like deep learning, to modify the training parameters is a significant obstacle to ensuring the convergence rate of multi-agent learning. Reducing communication overhead in advanced distributed optimization is another difficulty that needs to be addressed. This calls for the development of sophisticated communication techniques like intelligent encoding and precoding. When it comes to cache-enabled intelligent IoT networks, another obstacle is the complex allocation of resources among numerous agents. These resources include system training, caching, communication, and processing. To report on the state-of-the-art in research focusing on cache-enabled intelligent IoT networks and its applications, this book series aims to bring together scholars and practitioners from various domains into a shared platform.
Value: This book uses Artificial Intelligence’s ability to explain its actions to explore novel problems, domains of application, and methods for analysing human behaviour in natural settings. Through a computational intelligence system, these groups may reflect distinct ethnicities, genders, or even spatial or chronological differences like afternoon and evening consumers. Additionally, a computational intelligence system’s aggregate and personal fairness needs sometimes conflict. Additionally, regional moral and constitutional standards impact the bias of an AI-powered system. AI algorithms are being utilised to improve medical care operations, including intakes, prioritisation, patient insurance costs, and diagnostic forecasts. These uses may result in inequitable findings for certain demographic subgroups; hence, a statistically reliable fairness technique is needed. The next sections examine numerous ways to define and quantify fairness to fit an AI framework to a computational intelligence system.