Deep Learning From Spatio-Temporal Data Using Orthogonal Regularizaion Residual CNN for Air Prediction

Lei Zhang,Quansheng Guo,Dong Li

doi:10.1109/access.2020.2985657

Abstract

Air pollution is harmful to human health and restricts economic development, so predicting when and where air pollution will occur is a challenging and important issue, especially in fields of urban planning, factory production and human activities. In this paper, we propose a deep Spatio-Temporal Orthogonal Regularization Residual CNN (ST-OR-ResNet) for air prediction. Deep Convolutional Neural Network (CNN) is presented to capture the complex spatio-temporal relation of the dynamic biased meteorological data. Residual learning is designed to avoid unpredictable oscillations when training the network and verifying errors. For the issue of characteristic statistical migration and saddle point proliferation in deep network, the orthogonality regularizations are designed to stabilize the back-propagation errors, utilizing various advanced analytical tools such as restricted isometry property without extra hassle. We then benchmark their effects on public real-world datasets to demonstrate that ST-OR-ResNet has better predictive performance than the state-of-the-art methods.

Highlights

Nine out of ten people around the world breathe the polluted air, which kills seven million people a year, according to a report by WHO (World Health Organization) in 2018 [1]
Environmental authorities face some challenges of establishing ground stations to monitor the air pollution, which are expensive [3]: The associate editor coordinating the review of this manuscript and approving it for publication was Emre Koyuncu
A Deep Convolutional Neural Network (CNN) is designed to capture the complex correlation of spatio-temporal data and the edge effect in spatial distribution based on deep learning algorithm

Summary

INTRODUCTION

Nine out of ten people around the world breathe the polluted air, which kills seven million people a year, according to a report by WHO (World Health Organization) in 2018 [1]. L. Zhang et al.: Deep Learning From Spatio-Temporal Data Using Orthogonal Regularizaion Residual CNN for Air Prediction. A Deep CNN is designed to capture the complex correlation of spatio-temporal data and the edge effect in spatial distribution based on deep learning algorithm. Fang et al [10] proposed a satellite-based real-time adaptive method to predict the concentrations of PM2.5, and the reliability of the proposed algorithm was tested by combining meteorological factors, land utilization and other multi-source auxiliary data He and Huang [11] studied the spatio-temporal geographic weighted regression model to estimate AQI using AOD within a spatial resolution of 3km. Zhang et al [25] put forward spatio-temporal residual algorithm to predict the traffic flow, considering the time proximity, period and trend While these methods do predict the historical timestamps, they do not explicitly model the chronological order dependencies.

SPATIAL CONVERSION

EXPERIMENT AND PERFORMANCE ANALYSIS

DATA AND ENVIRONMENT

Findings

CONCLUSION AND FUTURE WORK