Live Demonstration: Real-time Analyses of Biosignals based on a Dedicated CMOS Configurable Deep Learning Engine

Abstract

Biosignals generated by human bodies contain valuable information about a person’s physical or psychological states. In recent years, machine-learning algorithms have significantly increased the accuracy and usefulness of biosignal analysis in areas such as disease diagnoses and treatments. To make these analyses more portable and accessible, we have designed and fabricated a dedicated processor named CODE, which supports machine-learning processing of various types of biosignals, including electroencephalography (EEG), electromyography (EMG), and electrocardiography (ECG), with high power efficiency and low latency. In this demonstration, we will show how the CODE chip processes biosignal data in real-time and show measurements of its power consumption and efficiency.