Fig. 1. Schematic showing the concept of a memristor-based RC system for neural activity analysis.

Inputs: Neural spikes collected from a firing neuron are directly used as inputs to excite the memristor. Reservoir: The reservoir space is further expanded by the concept of virtual nodes to help process complex temporal inputs. Readout: A simple ANN is used as readout layer of the reservoir to produce the final output (e.g. recognizing “Bursting” firing pattern in this example).