Intelligent adaptable systems: First order approach

Abstract

Future of communications inevitably calls for technologies that feature high-level flexibility, adaptability, and intelligence. Intelligent adaptable systems are particularly suitable for this mission. Majority of adaptable systems utilize neural networks. Artificial neural networks are systems with huge network-like interconnectivity. They are not programmed. Neural Networks gain valuable properties through the process of adaptation called learning. Presented intelligent adaptable system utilizes neural network technology. The system incorporates internal adaptability at several levels. It autonomously adapts its parameters and structure to the presented data. Externally, it appropriately manages its input-output interfaces. The system is able to select suitable training exemplars from the available amount of data in order to achieve the optimal learning performance. After training the system provides logical output format of the task. Introduced intelligent adaptable system consists of several modules. Principle and functionality of each module is described and illustratively demonstrated.