A Comparative Study: Toward an Effective Convolutional Neural Network Architecture for Sensor-Based Human Activity Recognition
A Comparative Study: Toward an Effective Convolutional Neural Network Architecture for Sensor-Based Human Activity Recognition
复制标题
DOI：
10.1109/access.2022.3152530
复制
发表时间：
2022
期刊：
影响因子：
3.9
通讯作者：
Zhongkai Zhao;Satoshi Kobayashi;Kazuma Kondo;Tatsuhito Hasegawa;M. Koshino
中科院分区：
文献类型：
--
作者：
Zhongkai Zhao;Satoshi Kobayashi;Kazuma Kondo;Tatsuhito Hasegawa;M. Koshino
研究了基于传感器数据的人体活动识别（HAR）特征提取方法。显著的特征提取能力是提高HAR精度的关键因素。近年来，深度学习方法被用于特征提取。在本文中，我们回顾了HAR中深度学习方法的研究，并讨论了适合的特征提取模型。首先，我们应用了各种卷积神经网络来阐明HAR的有效架构。随后，我们通过嵌入子模块开发了先进的模型，例如最近研究中经常采用的自注意和递归神经网络。在HASC、UCI和WISDM公共数据集上的对比实验表明，采用跨通道多尺度卷积变换的Inception-V3优于其他主干。通过嵌入子模块后的对比实验，子模块并不总是对精度产生积极的影响。与其他子模块相比，SENet具有积极的效果。我们得出结论，在应用子模块之前选择合适的骨干模型是必要的，在某些情况下，子模块是不必要的。
The feature extraction of human activity recognition (HAR) based on sensor data has been studied as a hand-crafted method. The significant feature extraction ability is a key factor in improving the accuracy of HAR. Recently, deep learning methods have been employed for feature extraction. In this paper, we review previous studies on deep learning methods in HAR and discuss suitable models for feature extraction. First, we applied various convolutional neural networks to clarify the effective architecture for HAR. Afterward, we developed advanced models by embedding submodules, such as self-attention and recurrent neural networks, often adopted in recent studies. Comparative experiments on HASC, UCI, and WISDM public datasets showed that Inception-V3, which used cross-channel multi-size convolution transformation, outperformed other backbones. Through comparative experiments after embedding submodules, submodules do not always have a positive effect on accuracy. Compared with other submodules, SENet has a positive effect. We conclude that it is essential to select an appropriate backbone model before applying the submodules, and submodules are unnecessary in some cases.