Fig. 6
Reproduced with permission from Ref. [30]. Copyright 2017, Springer Nature
Reservoir computing. a A illustration of reservoir computing consisting of input layer, reservoir, and output layer. The reservoir can nonlinearly transform incoming temporal inputs into reservoir states in a new feature space, which are then further learned and analyzed at the output layer. b A hardware implementation scheme of reservoir computing based on filament-free switching memristors. Reproduced with permission from Ref. [164]. Copyright 2021, Springer Nature. c A plot of read currents (reservoir states) in response to programming voltage pulses with varying input timings. d A digit pattern (left) and the reservoir states through five memristors with respect to the five input pulse trains corresponding to each row of the digit pattern. e, f An input pattern of the second-order nonlinear problem (left) and its solving result with a memristive RC system (right).