Certified Control Oriented Learning: A Robust Predictor-Based Approach

Rajiv Singh, Mario Sznaier

Abstract: We present a novel approach to the problem of learning the behavior of dynamical systems for the purpose of robust control design. The approach is centered around the derivation of stable predictors of potentially unstable systems and using them to identify plant models that can be ranked by their complexity (order) vs. empirical <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Nu</i> -gap value.