ELC-ECG: Efficient LSTM Cell for ECG Classification Based on Quantized Architecture

Abstract

Long Short-Term Memory (LSTM) is one of the most popular and effective Recurrent Neural Network (RNN) models used for sequence learning in applications such as ECG signal classification. Complex LSTMs could hardly be deployed on resource-limited bio-medical wearable devices due to the huge amount of computations and memory requirements. Binary LSTMs are introduced to cope with this problem. However, naive binarization leads to significant accuracy loss in ECG classification. In this paper, we propose an efficient LSTM cell along with a novel hardware architecture for ECG classification. By deploying 5-level binarized inputs and just 1level binarization for weights, output, and in-memory cell activations, the delay of one LSTM cell operation is reduced 50x with about 0.004% accuracy loss in comparison with full precision design of ECG classification.

Deep Learning-based ECG Classification on Raspberry PI using a Tensorflow Lite Model based on PTB-XL Dataset

“…Wavelets are also used for feature extraction, which could be used by the LSTM model [30]. Since binary LSTMs could run on the limited memory of wearable devices [31], their application is increasingly being explored. Numerous other models are being explored and adapted for the use case scenario.…”