A framework for cooperative adaptable information systems

Abstract

Both information systems and systems that automate complex control processes need to be able to adapt to new and possibly surprising situations, preferably without human intervention. These systems may not only need to control a domain, but also should be able to improve their own performance over time.This paper describes the initial implementation of a domain-independent Integrated Learning System (Ils), and one application, which, through its own experience, learns how to control the traffic in a telephone network. The issues involve coordinating distributed cooperating heterogeneous problem-solvers, combining various learning paradigms, and integrating different reasoning techniques.The Ils is a framework for integrating several heterogeneous learning agents that are written in different languages and run on different platforms; they cooperate to improve problem-solving performance. Ils also includes a central controller, called The Learning Coordinator (Tlc), which manages the control of flow and communication among the agents using a high-level communication protocol. The agents provide Tlc with expert advice. Tlc chooses which suggestion to adopt and performs the appropriate controls. At intervals, the agents can inspect the results of Tlc's actions and use this feedback to learn, improving the quality of their future advice.KeywordsTraffic ManagerTelephone NetworkTrunk GroupRecord MessageNetwork Traffic FlowThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.