Low Voltage

Low-voltage research focuses on mitigating the challenges and exploiting the benefits of operating systems and devices at reduced voltage levels. Current efforts concentrate on improving prediction accuracy of minimum operating voltage, enhancing robustness against bit errors induced by low voltage in applications like autonomous systems (using techniques like conformalized quantile regression and robust reinforcement learning), and developing accurate probabilistic forecasting models for low-voltage loads (employing neural networks and normalizing flows). These advancements are crucial for improving energy efficiency in various applications, from microchips to power grids, and for enabling more reliable and sustainable energy systems.