Cross-position activity recognition with stratified transfer learning
Cross-position activity recognition with stratified transfer learning
复制标题
通过分层迁移学习进行跨职位活动识别
DOI：
10.1016/j.pmcj.2019.04.004
复制
发表时间：
2019-07-01
影响因子：
4.3
通讯作者：
Yu, Han
中科院分区：
文献类型：
--
作者：
Chen, Yiqiang;Wang, Jindong;Yu, Han
人体活动识别（HAR）的目的是利用附着在不同身体部位的传感器来识别日常生活活动。HAR依赖于使用足够的活动数据训练的机器学习模型。然而，当来自某个身体位置（即目标域）的标签丢失时，如何利用来自其他位置（即源域）的数据来帮助识别该位置的活动？这个问题可以分为两个步骤。首先，当存在多个可用的源域时，通常难以选择与目标域最相似的源域。其次，对于选定的源域，我们需要在域之间进行准确的知识转移，以识别目标域上的活动。现有的方法只学习域间的全局距离，而忽略了局部属性。在本文中，我们提出了一个分层迁移学习（STL）框架来执行源域选择和活动迁移。STL基于我们提出的分层距离来捕获域的局部属性。STL由两个部分组成：1）分层领域选择（STL-SDS），可以选择与目标领域最相似的源领域; 2）分层活动转移（STL-SAT），能够进行准确的知识转移。在三个公共活动识别数据集上的大量实验证明了STL的优越性。(C)2019爱思唯尔B. V.保留所有权利。
Human activity recognition (HAR) aims to recognize the activities of daily living by utilizing the sensors attached to different body parts. HAR relies on the machine learning models trained using sufficient activity data. However, when the labels from a certain body position (i.e. target domain) are missing, how to leverage the data from other positions (i.e. source domain) to help recognize the activities of this position? This problem can be divided into two steps. Firstly, when there are several source domains available, it is often difficult to select the most similar source domain to the target domain. Secondly, with the selected source domain, we need to perform accurate knowledge transfer between domains in order to recognize the activities on the target domain. Existing methods only learn the global distance between domains while ignoring the local property. In this paper, we propose a Stratified Transfer Learning (STL) framework to perform both source domain selection and activity transfer. STL is based on our proposed Stratified distance to capture the local property of domains. STL consists of two components: 1) Stratified Domain Selection (STL-SDS), which can select the most similar source domain to the target domain; and 2) Stratified Activity Transfer (STL-SAT), which is able to perform accurate knowledge transfer. Extensive experiments on three public activity recognition datasets demonstrate the superiority of STL. (C) 2019 Elsevier B.V. All rights reserved.