Feature Representation and Data Augmentation for Human Activity Classification Based on Wearable IMU Sensor Data Using a Deep LSTM Neural Network.
Feature Representation and Data Augmentation for Human Activity Classification Based on Wearable IMU Sensor Data Using a Deep LSTM Neural Network.
复制标题
DOI：
10.3390/s18092892
复制
发表时间：
2018-08-31
期刊：
影响因子：
--
通讯作者：
Han DS
中科院分区：
文献类型：
--
作者：
Steven Eyobu O;Han DS
可穿戴惯性测量单元 (IMU) 传感器是运动数据采集的强大推动者。具体来说，在人类活动识别（HAR）中，从人体运动中收集的 IMU 传感器数据被分类组合以制定可用于学习人类活动的数据集。然而，从运动数据成功学习人类活动涉及设计和使用 IMU 传感器数据的正确特征表示和合适的分类器。此外，标记数据的稀缺是理解数据驱动学习模型性能过程中的一个障碍因素。为了应对这些挑战，本文有两个主要贡献：第一；通过使用原始 IMU 传感器数据，提出了一种基于频谱图的特征提取方法。其次，提出了特征空间中的数据增强集合来解决数据稀缺问题。在深度长期短期记忆 (LSTM) 神经网络架构上进行了性能测试，以探索特征表示和增强对活动识别准确性的影响。所提出的特征提取方法与数据增强集成相结合，在 HAR 中产生了最先进的精度结果。对每种增强方法进行性能评估，以显示对分类精度的影响。最后，除了使用我们自己的数据集之外，还针对加州大学欧文分校 (UCI) 公共在线 HAR 数据集对所提出的数据增强技术进行了评估，并在各种学习率下产生了最先进的准确性结果。
Wearable inertial measurement unit (IMU) sensors are powerful enablers for acquisition of motion data. Specifically, in human activity recognition (HAR), IMU sensor data collected from human motion are categorically combined to formulate datasets that can be used for learning human activities. However, successful learning of human activities from motion data involves the design and use of proper feature representations of IMU sensor data and suitable classifiers. Furthermore, the scarcity of labelled data is an impeding factor in the process of understanding the performance capabilities of data-driven learning models. To tackle these challenges, two primary contributions are in this article: first; by using raw IMU sensor data, a spectrogram-based feature extraction approach is proposed. Second, an ensemble of data augmentations in feature space is proposed to take care of the data scarcity problem. Performance tests were conducted on a deep long term short term memory (LSTM) neural network architecture to explore the influence of feature representations and the augmentations on activity recognition accuracy. The proposed feature extraction approach combined with the data augmentation ensemble produces state-of-the-art accuracy results in HAR. A performance evaluation of each augmentation approach is performed to show the influence on classification accuracy. Finally, in addition to using our own dataset, the proposed data augmentation technique is evaluated against the University of California, Irvine (UCI) public online HAR dataset and yields state-of-the-art accuracy results at various learning rates.
登录
查看更多内容
影响因子：
19.5
作者：
Laptev, I
通讯作者：
Laptev, I
DOI：
10.1109/tpami.2010.86
发表时间：
2011-04-01
影响因子：
23.6
作者：
Bulling, Andreas;Ward, Jamie A.;Troester, Gerhard
通讯作者：
Troester, Gerhard
影响因子：
11.8
作者：
Ding, Meng;Fan, Guoliang
通讯作者：
Fan, Guoliang
影响因子：
5
作者：
Chawla, NV;Bowyer, KW;Kegelmeyer, WP
通讯作者：
Kegelmeyer, WP
影响因子：
6.1
作者：
Kwolek, Bogdan;Kepski, Michal
通讯作者：
Kepski, Michal