Remarks on input to state stability of perturbed gradient flows, motivated by model-free feedback control learning

From MaRDI portal