Quantized Convolutional Neural Network toward Real-time Arrhythmia Detection in Edge Device
Muhammad Ilham Rizqyawan, Aris Munandar, M. Faizal Amri, Rio Korio Utoro, Agus Pratondo
Abstract
Automatic arrhythmia detection is one of the most researched areas in electrocardiography (ECG). Many methods have been proposed for the task using, not only the traditional machine learning but also deep learning algorithms. To build a real-time edge device, the algorithm should be fast but keep the accuracy high. In this paper, a convolutional neural network (CNN) model is quantized and tested to investigate its performance for the device. Results indicate that the CNN architecture is suitable for a real-time edge device. The speed is 58.8 times faster compared to the state-of-the-art methods.
Topics & Concepts
Convolutional neural networkComputer scienceEnhanced Data Rates for GSM EvolutionArtificial intelligenceDeep learningEdge deviceTask (project management)Edge detectionPattern recognition (psychology)Artificial neural networkImage (mathematics)Image processingEngineeringCloud computingOperating systemSystems engineeringECG Monitoring and AnalysisEEG and Brain-Computer InterfacesAnalog and Mixed-Signal Circuit Design