Learning Control-Relevant Koopman Representations from Input-Output Trajectorieswith Mohammad Khosravi1 versionW_tt3wbespv1 · current