Hybrid intelligent architecture for real time processing

The paper deals with the development of a hybrid intelligent architecture intended for work in complex real time client-server environment. The environment can be characterized as an environment with restricted time for making a response. Existing technologies cannot provide robust and intelligent solution, and traditional hybrid intelligent systems are not suitable, because of performance requirements. New architecture of a hybrid intelligent processing based on metaphor of “constructor” acting in an active environment is proposed for support of such real-time applications. The architecture provides integration of imperative high-level actions of an intelligent agent with work of active elements possessing own behavior and based on recurrent (neural) networks schemes. Methods for representation of the active environments are presented and operations on the schemes is developed. Application of developed architecture for building of servers are described