Decision making with neural networks

A neural network for distributed decision making is proposed. This decision making is a team effort, where a final decision-making neuron receives inputs from a set of consultant neurons. The consultant neurons range from 'lenient' to 'strict', depending on their firing thresholds, and have continuous outputs. The final decision-making neuron has binary output. The proposed network is envisioned as a portable tool for time-varying parameter identification and system fault detection, and may be used as part of an intelligent control scheme. As an application example, it is shown that, during on-line operation in conjunction with a fast parameter estimation algorithm, the decision-making network functions as an efficient detector of sudden changes.<
>