Resource and Energy Efficient Implementation of ECG Classifier Using Binarized CNN for Edge AI Devices
David Liang Tai Wong, Yongfu Li, Deepu John, Weng Khuen Ho, Chun-Huat Heng
Abstract
Wearable Artificial Intelligence-of-Things (AIoT) devices demand smart gadgets that are both resource and energy-efficient. In this paper, we explore efficient implementation of binary convolutional neural network employing function merging and block reuse techniques. The hardware implemented in field programmable gate array (FPGA) platform can classify ventricular beat in electrocardiogram achieving accuracy of 97.5%, sensitivity of 85.7%, specificity of 99.0%, precision of 92.3%, and F1-score of 88.9% while consuming only 10.5-μW of dynamic power dissipation.
Topics & Concepts
Field-programmable gate arrayComputer scienceConvolutional neural networkWearable computerEdge deviceArtificial neural networkArtificial intelligenceBinary numberReuseComputer hardwareClassifier (UML)Embedded systemPattern recognition (psychology)EngineeringArithmeticOperating systemMathematicsCloud computingWaste managementECG Monitoring and AnalysisEEG and Brain-Computer InterfacesAnalog and Mixed-Signal Circuit Design