Towards a Calculus of Echo State Networks

Reservoir computing is a recent trend in neural networks which uses dynamical perturbations in the phase space of a system to compute a desired target function. We show one can formulate an expectation of system performance in a simple model of reservoir computing called echo state networks. In contrast with previous theoretical frameworks, which uses annealed approximation, we calculate the exact optimal output weights as a function of the structure of the system and the properties of the input and the target function. Our calculation agrees with numerical simulations. To the best of our knowledge this work presents the first exact analytical solution to optimal output weights in echo state networks.