人工智能培训

搜索

深度学习论文:用于准确的人类活动识别的双残留网络(Dual Residual Network for Accurate Human Activity Recog

[复制链接]
zhq8008 发表于 2019-3-15 12:29:12 | 显示全部楼层 |阅读模式
zhq8008 2019-3-15 12:29:12 113 0 显示全部楼层
深度学习论文:用于准确的人类活动识别的双残留网络(Dual Residual Network for Accurate Human Activity Recognition)使用深度神经网络的人类活动识别(HAR)已经成为人机交互的热门话题。通过从大量传感器数据中学习,机器可以有效地识别人类自然活动。活动识别不仅是一个有趣的研究问题,而且还具有许多现实世界的实际应用。基于残余网络在实现自动化学习的高水平美学表现方面的成功,我们提出了一种新的\ textbf {D} ual \ textbf {R} esidual \ textbf {N} etwork,名为DRN。 DRN使用两个相同的路径框架来实现,所述路径框架包括(1)用于捕获空间特征的短时间窗口,以及(2)沿时间窗口,其用于捕获精细时间特征。通过减少其信道容量,可以使长时间窗口路径非常轻量级,但仍然能够学习用于活动识别的有用时间表示。在本文中,我们主要关注提出一种新模型来提高HAR的准确性。为了证明DRN模型的有效性,我们进行了广泛的实验,并与传统的识别方法(HC,CBH,CBS)和基于学习的方法(AE,MLP,CNN,LSTM,Hybrid,ResNet)进行了比较。基准数据集(OPPORTUNITY,UniMiB-SHAR)由ourexperiments采用。我们的实验结果表明,我们的模型可以通过可穿戴数据集识别人类活动。我们讨论了网络参数对性能的影响,以提供有关其优化的见解。
Human Activity Recognition (HAR) using deep neural network has become a hottopic in human-computer interaction.Machine can effectively identify humannaturalistic activities by learning from a large collection of sensor data.Activity recognition is not only an interesting research problem, but also hasmany real-world practical applications.Based on the success of residualnetworks in achieving a high level of aesthetic representation of the automaticlearning, we propose a novel \textbf{D}ual \textbf{R}esidual \textbf{N}etwork,named DRN.DRN is implemented using two identical path frameworks consisting of(1) a short time window, which is used to capture spatial features, and (2) along time window, which is used to capture fine temporal features.The longtime window path can be made very lightweight by reducing its channel capacity,yet still being able to learn useful temporal representations for activityrecognition.In this paper, we mainly focus on proposing a new model to improvethe accuracy of HAR.In order to demonstrate the effectiveness of DRN model, wecarried out extensive experiments and compared with conventional recognitionmethods (HC, CBH, CBS) and learning-based methods (AE, MLP, CNN, LSTM, Hybrid,ResNet).The benchmark datasets (OPPORTUNITY, UniMiB-SHAR) were adopted by ourexperiments.Results from our experiments show that our model is effective inrecognizing human activities via wearable datasets.We discuss the influence ofnetworks parameters on performance to provide insights about its optimization.深度学习论文:用于准确的人类活动识别的双残留网络(Dual Residual Network for Accurate Human Activity Recognition) Ul083A8Djl88pIP0.jpg
URL地址:https://arxiv.org/abs/1903.05359     ----pdf下载地址:https://arxiv.org/pdf/1903.05359    ----深度学习论文:用于准确的人类活动识别的双残留网络(Dual Residual Network for Accurate Human Activity Recognition)
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

zhq8008当前离线
新手上路

查看:113 | 回复:0

快速回复 返回顶部 返回列表