HAR-Net: Fusing Deep Representation and Hand-crafted Features for Human Activity Recognition
HAR-Net: Fusing Deep Representation and Hand-crafted Features for Human Activity Recognition
复制标题
HAR-Net：融合深度表示和手工特征以进行人类活动识别
DOI：
10.1007/978-981-13-7123-3_4
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Jindong Han
中科院分区：
文献类型：
--
作者：
Mingtao Dong;Jindong Han
可穿戴计算和情境感知是近年来人工智能领域的研究热点。最有吸引力和最具挑战性的应用之一是利用智能手机的人类活动识别（HAR）。基于支持向量机的传统 HAR 依赖于手动提取特征。由于人类对提取特征的看法不全面，这种方法在预测中非常耗时和耗能。随着深度学习的兴起，人工智能正在向成熟技术迈进。本文提出了一种基于深度学习的新方法 HAR-Net 来解决 HAR 问题。该研究使用了 Android 智能手机中的陀螺仪和加速度传感器收集的数据。 HAR-Net 融合了手工制作的特征和从卷积神经网络中提取的高级特征来进行预测。事实证明，所提出方法的性能高于原始MC-SVM方法。 UCI数据集上的实验结果表明，融合两种特征可以弥补传统特征工程和深度学习技术的不足。
Wearable computing and context awareness are the focuses of study in the field of artificial intelligence recently. One of the most appealing as well as challenging applications is the Human Activity Recognition (HAR) utilizing smart phones. Conventional HAR based on Support Vector Machine relies on manually extracted features. This approach is time and energy consuming in prediction due to the partial view toward which features to be extracted by human. With the rise of deep learning, artificial intelligence has been making progress toward being a mature technology. This paper proposes a new approach based on deep learning called HAR-Net to address the HAR issue. The study used the data collected by gyroscopes and acceleration sensors in android smart phones. The HAR-Net fusing the hand-crafted features and high-level features extracted from convolutional neural network to make prediction. The performance of the proposed method was proved to be higher than the original MC-SVM approach. The experimental results on the UCI dataset demonstrate that fusing the two kinds of features can make up for the shortage of traditional feature engineering and deep learning techniques.