三轴加速度传感器数据去识别各种行为时会出现模型识别能力低下的问题.其他的错误发生在 upstairs和walking之间.行为模态的相似性使得upstairs的l2条数据被错误地分成了walking. 3结论 (1)针对人体行为识别研究中资源有限问题，本文提出了一种轻量型NCA-L-SCNs人体行为 识别学习模型： (2)NCA特征选择方法能够提高人体行为识别特征集的可分性和轻量性，进而达到提高模型 的识别精度和降低模型建模过程计算复杂度： (3)使用L2正则化技术解决SCNs由于隐含层节点过多导致的过拟合问题，增强SCNs模型结 构的紧致性，进而提高模型的泛化性和轻量性： (4)在UCI HAR特征集上的实验结果表明，在可接受的时间内，本文所提NCA-L2-SCNs学习 模型相比于其他模型具有更好的识别精度，且计算复杂度更低.因此，模型更轻量 参考文献 [1]Mukherjee D,Mondal R,Singh P K,et al.EnsemConvNet:a deep learning approach for human activity recognition using smartphone sensors for healthcare applications.Multimedia Tools and Applieations,2020,79(Part 2):1-28. [2]Zhuang Z D,Yang X.Sport-Related Human Activity Detection and Recognition Using a Smartwatch.Sensors (Basel,Switzerland),2019,19(22). [3]Ibrahim AA,Ame K,Ganer H,et al.Inertial sensor-based gait parameters reflect patient-reported fatigue in multiple sclerosis.Journal of NeuroEngineering and Rehabilitation,202017(1) [4]Hassan MM,Ullah S,Hossain M S,et al.An end-to-end deep-learning model for human activity recognition from highly sparse body sensor data in Internet of Medical Things environment.The Journal of Supercomputing,2020(4). [5]Igwe O M,Wang Y,Giakos G C,et al.Human activity recognition in smart environments employing margin setting algorithm.Journal of Ambient Intelligence and Humanized Computing,2020. [6]Fang H,Tang P.Si H.Feature Selections Using Minimal Redundancy Maximal Relevance Algorithm for Human Activity Recognition in Smart Home Environments.Journal of Healthcare Engineering,2020,2020(1):1-13. [7]Heinrich K M,Spencer V,Fehl N,ta Mission essential fitness:comparison of functional circuit training to traditional Army physical training for active duty military.Military Medicine,2012,177(10):1125-30. [8]Foerster F,Smeja M,Fahrenbef Detection of posture and motion by accelerometry:a validation study in ambulatory monitoring.Comput Hum Behay:Computers in Human Behavior,1999,15(5):571-583. [9]Bharti P.Complex activity recognition with multi-modal multi-positional body sensing.Journal of Biometrics Biostatistics,201708(5). [10]Chen Z,Jiang C.Xie L.A Novel Ensemble ELM for Human Activity Recognition Using Smartphone Sensors.IEEE transactions on industrial informatics,201915(5):691-2699. [11]Abidine M B./Fergani B,Menhour I.Activity Recognition from Smartphones Using Hybrid Classifier PCA-SVM- HMM//2019 International Conference on Wireless Networks and Mobile Communications (WINCOM).2019. [12]Mohammad Y,Matsumoto K,Hoashi K.Selecting Orientation-Insensitive Features for Activity Recognition from Accelerometers.IEICE Transactions on Information and Systems,2019,E102.D(1):104-115. [13]Sansano E,Montoliu R,Fernandez S B.A study of deep neural networks for human activity recognition.Computational Intelligence,2020,36(6). [14]Abbaspour S,Fotouhi F,Sedaghatbaf A,et al.A Comparative Analysis of Hybrid Deep Learning Models for Human Activity Recognition.Sensors,2020,20(5707). [15]Zou Q,Wang Y,Wang Q,et al.Deep Learning-Based Gait Recognition Using Smartphones in the Wild.IEEE三轴加速度传感器数据去识别各种行为时会出现模型识别能力低下的问题 . 其他的错误发生在 upstairs 和 walking 之间. 行为模态的相似性使得 upstairs 的 12 条数据被错误地分成了 walking. 3 结论 (1) 针对人体行为识别研究中资源有限问题，本文提出了一种轻量型 NCA-L2-SCNs 人体行为 识别学习模型； (2) NCA 特征选择方法能够提高人体行为识别特征集的可分性和轻量性，进而达到提高模型 的识别精度和降低模型建模过程计算复杂度； (3) 使用 L2正则化技术解决 SCNs 由于隐含层节点过多导致的过拟合问题，增强 SCNs 模型结 构的紧致性，进而提高模型的泛化性和轻量性； (4) 在 UCI HAR 特征集上的实验结果表明，在可接受的时间内，本文所提 NCA-L2-SCNs 学习 模型相比于其他模型具有更好的识别精度，且计算复杂度更低. 因此，模型更加轻量. 参 考 文 献 [1] Mukherjee D, Mondal R, Singh P K, et al. EnsemConvNet: a deep learning approach for human activity recognition using smartphone sensors for healthcare applications. Multimedia Tools and Applications, 2020, 79(Part 2):1-28. [2] Zhuang Z D, Yang X. Sport-Related Human Activity Detection and Recognition Using a Smartwatch. Sensors (Basel,Switzerland), 2019, 19(22). [3] Ibrahim A A, Arne K, Ganer H, et al. Inertial sensor-based gait parameters reflect patient-reported fatigue in multiple sclerosis. Journal of NeuroEngineering and Rehabilitation, 2020, 17(1). [4] Hassan M M, Ullah S, Hossain M S, et al. An end-to-end deep learning model for human activity recognition from highly sparse body sensor data in Internet of Medical Things environment. The Journal of Supercomputing, 2020(4). [5] Igwe O M, Wang Y, Giakos G C, et al. Human activity recognition in smart environments employing margin setting algorithm. Journal of Ambient Intelligence and Humanized Computing, 2020. [6] Fang H, Tang P, Si H. Feature Selections Using Minimal Redundancy Maximal Relevance Algorithm for Human Activity Recognition in Smart Home Environments. Journal of Healthcare Engineering, 2020, 2020(1):1-13. [7] Heinrich K M, Spencer V, Fehl N, et al. Mission essential fitness: comparison of functional circuit training to traditional Army physical training for active duty military. Military Medicine, 2012, 177(10):1125-30. [8] Foerster F, Smeja M, Fahrenberg J. Detection of posture and motion by accelerometry: a validation study in ambulatory monitoring. Comput Hum Behav. Computers in Human Behavior, 1999, 15(5):571-583. [9] Bharti P. Complex activity recognition with multi-modal multi-positional body sensing. Journal of Biometrics & Biostatistics, 2017, 08(5). [10] Chen Z, Jiang C, Xie L. A Novel Ensemble ELM for Human Activity Recognition Using Smartphone Sensors. IEEE transactions on industrial informatics, 2019, 15(5):2691-2699. [11] Abidine M B, Fergani B, Menhour I. Activity Recognition from Smartphones Using Hybrid Classifier PCA-SVM￾HMM// 2019 International Conference on Wireless Networks and Mobile Communications (WINCOM). 2019. [12] Mohammad Y, Matsumoto K, Hoashi K. Selecting Orientation-Insensitive Features for Activity Recognition from Accelerometers. IEICE Transactions on Information and Systems, 2019, E102.D(1):104-115. [13] Sansano E, Montoliu R, Fernández S B. A study of deep neural networks for human activity recognition. Computational Intelligence, 2020, 36(6). [14] Abbaspour S, Fotouhi F, Sedaghatbaf A, et al. A Comparative Analysis of Hybrid Deep Learning Models for Human Activity Recognition. Sensors, 2020, 20(5707). [15] Zou Q, Wang Y, Wang Q, et al. Deep Learning-Based Gait Recognition Using Smartphones in the Wild. IEEE 录用稿件，非最终出版稿