Understanding and Improving Deep Neural Network for Activity Recognition

Abstract

Activity recognition has become a popular research branch in the field of pervasive computing in recent years. A large number of experiments can be obtained that activity sensor-based data's characteristic in activity recognition is variety, volume, and velocity. Deep learning technology, together with its various models, is one of the most effective ways of working on activity data. Nevertheless, there is no clear understanding of why it performs so well or how to make it more effective. In order to solve this problem, first, we applied convolution neural network on Human Activity Recognition Using Smart phones Data Set. Second, we realized the visualization of the sensor-based activity's data features extracted from the neural network. Then we had in-depth analysis of the visualization of features, explored the relationship between activity and features, and analyzed how Neural Networks identify activity based on these features. After that, we extracted the significant features related to the activities and sent the features to the DNN-based fusion model, which improved the classification rate to 96.1%. This is the first work to our knowledge that visualizes abstract sensor-based activity data features. Based on the results, the method proposed in the paper promises to realize the accurate classification of sensorbased activity recognition.

RETRACTED ARTICLE: Human activity recognition with deep learning: overview, challenges and possibilities

“…This is, however, completely insufficient, as deep models can still encode noise such as unrelevant methods [150]. Several scientists [151], [152] have demonstrated the versatility of neural networks. After the authors discover their connections with the behaviors of the simulation, salient characteristics are sent to the following models [152].…”

“…Several scientists [151], [152] have demonstrated the versatility of neural networks. After the authors discover their connections with the behaviors of the simulation, salient characteristics are sent to the following models [152]. Nutter et al [153] converted sensory data into images, allowing for more straightforward interpretability of visualization resources for sensory data.…”

Human Activity Recognition Algorithm Based on One-Dimensional Convolutional Neural Network

“…Singh et al [19] demonstrated the use of the CNN and a comparison of results, which has been performed with Long Short Term Memory (LSTM). Xue et al [20] realized the visualization of the sensor-based activity's data features extracted from the neural network and sent the features to the DNN-based fusion model, with a reported accuracy of 96.1%.…”

Wearable Accelerometer Layout Optimization for Activity Recognition Based on Swarm Intelligence and User Preference

“…Therefore, we conducted an experiment that applied the MSMS-DPSO to several different classification approaches. Since we selected the SVM as the chief classifier for our study, the compared methods are (a) Random Forest (30 estimators), (b) baseline CNN classifier, (c) DNN+LSTM from [1], and (d) DeepCNN [3]. Methods (c) and (d) maintained the network structure described in the original papers.…”