Low-Power Display System Enabled by Combining Oxide-Semiconductor and Neural-Network Technologies

Oxide semiconductor field-effect transistors (OSFETs) are characterized by their extremely low off-state current. By fabricating both a display and peripheral large-scale integrated circuits with OSFETs, the components can be driven in a normally off manner, thus enabling a display system with low power consumption. We explore herein the feasibility of a more energy-efficient display system, which is achieved by predicting changes in image data and thus more efficiently controlling the power-gate timing. This is done by using an artificial neural network.