閱讀全文 | |
篇名 |
End-to-End Deep Learning-Based Human Activity Recognition Using Channel State Information
|
---|---|
並列篇名 | End-to-End Deep Learning-Based Human Activity Recognition Using Channel State Information |
作者 | Chaur-Heh Hsieh、Jen-Yang Chen、Chung-Ming Kuo、Ping Wang |
英文摘要 | The automatic recognition of human activities in the house using Channel State Information (CSI) has received wide attention due to the potential use of wide range of intelligent services. Two-dimensional Convolutional Neural Network (2D-CNN) is one of the most popular approaches for human activity recognition (HAR). This method first applies signal transform to convert a time-series CSI signal into a 2D image, and then uses the image to train a complex 2D-CNN model. In this paper we will present simple deep neural networks including multi-layer perceptron (MLP) and onedimensional Convolutional Neural Network (1D-CNN) for HAR. Our proposed networks are fully end-to-end automatic learning from feature extraction/selection and classification, and do not require extra signal transform and denoising. Experimental results indicate that the proposed networks not only achieve much better recognition performance but reduces the network complexity significantly, as compared to the existing methods. |
起訖頁 | 271-281 |
關鍵詞 | Human activity recognition、Deep learning、Multi-Layer Perceptron (MLP)、Convolutional Neural Network (CNN)、Channel State Information (CSI) |
刊名 | 網際網路技術學刊 |
期數 | 202103 (22:2期) |
出版單位 | 台灣學術網路管理委員會 |
DOI |
|
QR Code | |
該期刊 上一篇
| Multi-group Flower Pollination Algorithm Based on Novel Communication Strategies |
該期刊 下一篇
| Street-level Landmark Mining Algorithm Based on Radar Search |