Advisor-enhanced real-time system control

The paper describes a hybrid signature-table controller that uses advisor logic to enhance its decision making and statistical-inference learning processes. The automaton reacts to changes in real state space with rapid-fire control decisions and uses timeouts or failures to initiate learning interludes. The post-processing algorithm causes the system to learn by adapting values in the control matrix with a process that utilizes computed decision strengths for each state and those of its advising neighbors. The familiar inverted pendulum system is used to highlight the effects of advisor logic on overall system performance.