Leaky FinFET for Reservoir Computing with Temporal Signal Processing

Reservoir computing can greatly reduce the hardware andtrainingcosts of recurrent neural networks with temporal data processing.To implement reservoir computing in a hardware form, physical reservoirstransforming sequential inputs into a high-dimensional feature spaceare necessary. In this work, a physical reservoir with a leaky fin-shapedfield-effect transistor (L-FinFET) is demonstrated by the positiveuse of a short-term memory property arising from the absence of anenergy barrier to suppress the tunneling current. Nevertheless, theL-FinFET reservoir does not lose its multiple memory states. The L-FinFETreservoir consumes very low power when encoding temporal inputs becausethe gate serves as an enabler of the write operation, even in theoff-state, due to its physical insulation from the channel. In addition,the small footprint area arising from the scalability of the FinFETdue to its multiple-gate structure is advantageous for reducing thechip size. After the experimental proof of 4-bit reservoir operationswith 16 states for temporal signal processing, handwritten digitsin the Modified National Institute of Standards and Technology datasetare classified by reservoir computing.