Shallow Convolutional Neural Networks for Human Activity Recognition Using Wearable Sensors

Due to rapid development of sensor technology, human activity recognition (HAR) using wearable inertial sensors has recently become a new research hotspot. Deep learning, especially convolutional neural network (CNN) that can automatically learn intricate activity features have gained a lot of attention in ubiquitous HAR task. Most existing CNNs process sensor input by extracting channel-wise features, and the information from each channel can be separately propagated in a hierarchical way from lower layers to higher layers. As a result, they typically overlook information exchange among channels within the same layer. In this article, we first propose a shallow CNN that considers cross-channel communication in HAR scenario, where all channels in the same layer have a comprehensive interaction to capture more discriminative features of sensor input. One channel can communicate with all other channels by graph neural network to remove redundant information accumulated among channels, which is more beneficial for deploying lightweight deep models. Extensive experiments are conducted on multiple benchmark HAR datasets, namely UCI-HAR, OPPORTUNITY, PAMAP2 and UniMib-SHAR, which indicates that the proposed method enables shallower CNNs to aggregate more useful information, and surpasses baseline deep networks and other competitive methods. The inference speed is evaluated via deploying the HAR systems on an embedded system.

I. Introduction

With rapid development of the Internet-of-Things and sensor technology, human activity recognition (HAR) using wearable inertial sensors has become a new research hotspot due to its extensive use in a large variety of application domains such as health-care [1], sports tracking [2], [3], fitness, game console design [4] and smart homes [5]. Deep learning [6], [7] has gained a lot of attention in sensor-based HAR scenario. Especially, convolutional neural networks (CNNs) have started delivering their advantages over feature learning and achieved state-of-the-art performance for HAR [8], [9]. Traditionally, various methods from the field of signal processing [10], [11] have been widely leveraged to distill collected sensor data, which requires domain-specific expert knowledge to process raw data. Statistical and machine learning models are then trained on the version of processed data [12]. That is to say, feature engineering is required to fit a model, which is expensive and not scalable. CNNs are capable of performing automatic feature learning, which significantly surpasses models fit on hand-crafted domain-specific features. Ideally, CNNs with automatic feature extraction provide the ability to learn features from raw sensor data with little pre-processing involved in feature engineering.