Vision-based deep Q-learning network models to predict particulate matter concentration levels using temporal digital image data

Sung Hwan Kim, Se Hee Jung, Seong Min Yang, Ji Seong Han, Byung Yong Lee, Jae Hwa Lee, Sung Won Han, Sidike Paheding

Research output: Contribution to journalArticle

Abstract

Particulate matter (PM) has been revealed to have detrimental effects on public health, social economy, agriculture, and so forth. Thus, it became one of the major concerns in terms of a factor that can reduce "quality of life" over East Asia, where the concentration is significantly high. In this regard, it is imperative to develop affordable and efficient prediction models to monitor real-time changes in PM concentration levels using digital images, which are readily available for many individuals (e.g., via mobile phone). Previous studies (i.e., DeepHaze) were limited in scope to priorly collected data and thereby less practical in providing real-time information (i.e., undermined interprediction). This drawback led us to hardly capture drastic changes caused by weather or regions of interests. To address this challenge, we propose a new method called Deep Q-haze, whose inference scheme is built on an online learning-based method in collaboration with reinforcement learning and deep learning (i.e., Deep Q-learning), making it possible to improve testing accuracy and model flexibility in virtue of real-time basis inference. Taking into account various experiment scenarios, the proposed method learns a binary decision rule on the basis of video sequences to predict, in real time, whether the level of PM10 (particles smaller than 10 in aerodynamic diameter) concentration is harmful (>80μg/m3) or not. The proposed model shows superior accuracy compared to existing algorithms. Deep Q-haze effectively accounts for unexpected environmental changes in essence (e.g., weather) and facilitates monitoring of real-time PM10 concentration levels, showing implications for better understanding of characteristics of airborne particles.

Original languageEnglish
Article number9673047
JournalJournal of Sensors
Volume2019
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

particulates
learning
haze
inference
weather
Reinforcement learning
Public health
Mobile phones
Agriculture
Aerodynamics
public health
agriculture
economy
reinforcement
aerodynamics
Monitoring
Testing
flexibility
Experiments
predictions

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Instrumentation
  • Electrical and Electronic Engineering

Cite this

Vision-based deep Q-learning network models to predict particulate matter concentration levels using temporal digital image data. / Kim, Sung Hwan; Jung, Se Hee; Yang, Seong Min; Han, Ji Seong; Lee, Byung Yong; Lee, Jae Hwa; Han, Sung Won; Paheding, Sidike.

In: Journal of Sensors, Vol. 2019, 9673047, 01.01.2019.

Research output: Contribution to journalArticle

Kim, Sung Hwan ; Jung, Se Hee ; Yang, Seong Min ; Han, Ji Seong ; Lee, Byung Yong ; Lee, Jae Hwa ; Han, Sung Won ; Paheding, Sidike. / Vision-based deep Q-learning network models to predict particulate matter concentration levels using temporal digital image data. In: Journal of Sensors. 2019 ; Vol. 2019.
@article{214674ba48264b32a2238172d81948c6,
title = "Vision-based deep Q-learning network models to predict particulate matter concentration levels using temporal digital image data",
abstract = "Particulate matter (PM) has been revealed to have detrimental effects on public health, social economy, agriculture, and so forth. Thus, it became one of the major concerns in terms of a factor that can reduce {"}quality of life{"} over East Asia, where the concentration is significantly high. In this regard, it is imperative to develop affordable and efficient prediction models to monitor real-time changes in PM concentration levels using digital images, which are readily available for many individuals (e.g., via mobile phone). Previous studies (i.e., DeepHaze) were limited in scope to priorly collected data and thereby less practical in providing real-time information (i.e., undermined interprediction). This drawback led us to hardly capture drastic changes caused by weather or regions of interests. To address this challenge, we propose a new method called Deep Q-haze, whose inference scheme is built on an online learning-based method in collaboration with reinforcement learning and deep learning (i.e., Deep Q-learning), making it possible to improve testing accuracy and model flexibility in virtue of real-time basis inference. Taking into account various experiment scenarios, the proposed method learns a binary decision rule on the basis of video sequences to predict, in real time, whether the level of PM10 (particles smaller than 10 in aerodynamic diameter) concentration is harmful (>80μg/m3) or not. The proposed model shows superior accuracy compared to existing algorithms. Deep Q-haze effectively accounts for unexpected environmental changes in essence (e.g., weather) and facilitates monitoring of real-time PM10 concentration levels, showing implications for better understanding of characteristics of airborne particles.",
author = "Kim, {Sung Hwan} and Jung, {Se Hee} and Yang, {Seong Min} and Han, {Ji Seong} and Lee, {Byung Yong} and Lee, {Jae Hwa} and Han, {Sung Won} and Sidike Paheding",
year = "2019",
month = "1",
day = "1",
doi = "10.1155/2019/9673047",
language = "English",
volume = "2019",
journal = "Journal of Sensors",
issn = "1687-725X",
publisher = "Hindawi Publishing Corporation",

}

TY - JOUR

T1 - Vision-based deep Q-learning network models to predict particulate matter concentration levels using temporal digital image data

AU - Kim, Sung Hwan

AU - Jung, Se Hee

AU - Yang, Seong Min

AU - Han, Ji Seong

AU - Lee, Byung Yong

AU - Lee, Jae Hwa

AU - Han, Sung Won

AU - Paheding, Sidike

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Particulate matter (PM) has been revealed to have detrimental effects on public health, social economy, agriculture, and so forth. Thus, it became one of the major concerns in terms of a factor that can reduce "quality of life" over East Asia, where the concentration is significantly high. In this regard, it is imperative to develop affordable and efficient prediction models to monitor real-time changes in PM concentration levels using digital images, which are readily available for many individuals (e.g., via mobile phone). Previous studies (i.e., DeepHaze) were limited in scope to priorly collected data and thereby less practical in providing real-time information (i.e., undermined interprediction). This drawback led us to hardly capture drastic changes caused by weather or regions of interests. To address this challenge, we propose a new method called Deep Q-haze, whose inference scheme is built on an online learning-based method in collaboration with reinforcement learning and deep learning (i.e., Deep Q-learning), making it possible to improve testing accuracy and model flexibility in virtue of real-time basis inference. Taking into account various experiment scenarios, the proposed method learns a binary decision rule on the basis of video sequences to predict, in real time, whether the level of PM10 (particles smaller than 10 in aerodynamic diameter) concentration is harmful (>80μg/m3) or not. The proposed model shows superior accuracy compared to existing algorithms. Deep Q-haze effectively accounts for unexpected environmental changes in essence (e.g., weather) and facilitates monitoring of real-time PM10 concentration levels, showing implications for better understanding of characteristics of airborne particles.

AB - Particulate matter (PM) has been revealed to have detrimental effects on public health, social economy, agriculture, and so forth. Thus, it became one of the major concerns in terms of a factor that can reduce "quality of life" over East Asia, where the concentration is significantly high. In this regard, it is imperative to develop affordable and efficient prediction models to monitor real-time changes in PM concentration levels using digital images, which are readily available for many individuals (e.g., via mobile phone). Previous studies (i.e., DeepHaze) were limited in scope to priorly collected data and thereby less practical in providing real-time information (i.e., undermined interprediction). This drawback led us to hardly capture drastic changes caused by weather or regions of interests. To address this challenge, we propose a new method called Deep Q-haze, whose inference scheme is built on an online learning-based method in collaboration with reinforcement learning and deep learning (i.e., Deep Q-learning), making it possible to improve testing accuracy and model flexibility in virtue of real-time basis inference. Taking into account various experiment scenarios, the proposed method learns a binary decision rule on the basis of video sequences to predict, in real time, whether the level of PM10 (particles smaller than 10 in aerodynamic diameter) concentration is harmful (>80μg/m3) or not. The proposed model shows superior accuracy compared to existing algorithms. Deep Q-haze effectively accounts for unexpected environmental changes in essence (e.g., weather) and facilitates monitoring of real-time PM10 concentration levels, showing implications for better understanding of characteristics of airborne particles.

UR - http://www.scopus.com/inward/record.url?scp=85068865788&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85068865788&partnerID=8YFLogxK

U2 - 10.1155/2019/9673047

DO - 10.1155/2019/9673047

M3 - Article

AN - SCOPUS:85068865788

VL - 2019

JO - Journal of Sensors

JF - Journal of Sensors

SN - 1687-725X

M1 - 9673047

ER -