Air Pollution Prediction Using Long Short-Term Memory (LSTM) and Deep Autoencoder (DAE) Models

Title : Air Pollution Prediction Using Long Short-Term Memory (LSTM) and Deep Autoencoder (DAE) Models
Journal : Sustainability
Authors : Thanongsak Xayasouk, HwaMin Lee, Giyeol Lee
These authors contributed equally to this work : Thanongsak Xayasouk, HwaMin Lee
Corresponding author : Giyeol Lee
DOI : https://doi.org/10.3390/su12062570

Abstract
Many countries worldwide have poor air quality due to the emission of particulate matter (i.e., PM10 and PM2.5), which has led to concerns about human health impacts in urban areas. In this study, we developed models to predict fine PM concentrations using long short-term memory (LSTM) and deep autoencoder (DAE) methods, and compared the model results in terms of root mean square error (RMSE). We applied the models to hourly air quality data from 25 stations in Seoul, South Korea, for the period from 1 January 2015, to 31 December 2018. Fine PM concentrations were predicted for the 10 days following this period, at an optimal learning rate of 0.01 for 100 epochs with batch sizes of 32 for LSTM model, and DAEs model performed best with batch size 64. The proposed models effectively predicted fine PM concentrations, with the LSTM model showing slightly better performance. With our forecasting model, it is possible to give reliable fine dust prediction information for the area where the user is located.

Acknowledgments
This research was supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2020-2015-0-00403)supervised by the IITP(Institute for Information &communications Technology Planning &Evaluation) and by Soonchunhyang Research Fund.

A Hybrid Deep Learning Model to Forecast Particulate Matter Concentration Levels in Seoul, South Korea

Title : A Hybrid Deep Learning Model to Forecast Particulate Matter Concentration Levels in Seoul, South Korea
Journal : Atmosphere
Authors : Guang Yang, HwaMin Lee, Giyeol Lee
These authors contributed equally to this work : Guang Yang, HwaMin Lee
Corresponding author : Giyeol Lee
DOI : https://doi.org/10.3390/atmos11040348

Abstract
Both long- and short-term exposure to high concentrations of airborne particulate matter (PM) severely affect human health. Many countries now regulate PM concentrations. Early-warning systems based on PM concentration levels are urgently required to allow countermeasures to reduce harm and loss. Previous studies sought to establish accurate, efficient predictive models. Many machine-learning methods are used for air pollution forecasting. The long short-term memory and gated recurrent unit methods, typical deep-learning methods, reliably predict PM levels with some limitations. In this paper, the authors proposed novel hybrid models to combine the strength of two types of deep learning methods. Moreover, the authors compare hybrid deep-learning methods (convolutional neural network (CNN)—long short-term memory (LSTM) and CNN—gated recurrent unit (GRU)) with several stand-alone methods (LSTM, GRU) in terms of predicting PM concentrations in 39 stations in Seoul. Hourly air pollution data and meteorological data from January 2015 to December 2018 was used for these training models. The results of the experiment confirmed that the proposed prediction model could predict the PM concentrations for the next 7 days. Hybrid models outperformed single models in five areas selected randomly with the lowest root mean square error (RMSE) and mean absolute error (MAE) values for both PM10 and PM2.5. The error rate for PM10 prediction in Gangnam with RMSE is 1.688, and MAE is 1.161. For hybrid models, the CNN–GRU better-predicted PM10 for all stations selected, while the CNN–LSTM model performed better on predicting PM2.5.

Acknowledgments
This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (NRF-2017R1A2B4010570) and Soonchunhyang University Research Fund.

위험상황감지 기능을 갖는 유아용 가방 및 이를 포함하는 위험상황감지 시스템

발명의 명칭 : 위험상황감지 기능을 갖는 유아용 가방 및 이를 포함하는 위험상황감지 시스템
출원번호 : 10-2018-0161132
출원일자 : 2018년 12월 13일
등록번호 : 10-2089874
등록일자 : 2020년 3월 10일
발명자 : 이화민, 이윤신
특허권자 : 순천향대학교 산학협력단
법적상태 : 등록
요약
본 발명은 위험상황감지 기능을 갖는 유아용 가방 및 이를 포함하는 위험상황감지 시스템을 개시한다. 본 발명의 일 측면에 따른 위험상황감지 기능을 갖는 유아용 가방은, 유아의 현재 위치 정보를 생성하는 위치 정보 획득부; 유아가 현재 위치한 장소의 온도 정보를 생성하는 온도 정보 획득부; 유아가 위험 상황에 처한 경우, 이를 알리기 위한 버튼부; 유아에 의해 상기 버튼부가 눌려지거나 또는 상기 온도 정보 획득부가 획득한 온도 정보가 미리 설정된 임계 온도를 벗어나는 경우 상기 유아가 위험한 상황에 처해 있다고 판단하는 위험 상황 감지부; 상기 위험 상황 감지부에 의해 유아가 위험한 상황에 처해 있다고 판단하는 경우 소리 또는 빛을 외부로 방송하는 알람부; 및 상기 위험 상황 감지부에 의해 유아가 위험한 상황에 처해 있다고 판단하는 경우, 상기 위치 정보 및 온도 정보를 포함하는 위험 상황 정보를 통신부를 통해 부모가 휴대한 단말로 전송하는 제어부;를 포함한다.

연구사사
이 발명을 지원한 국가연구개발사업
과제고유번호 20141007200051001
부처명 과학기술정보통신부
연구관리전문기관 정보통신기술진흥센터(IITP)
연구사업명 대학ICT연구센터육성지원사업
연구과제명 웰니스 삶을 위한 WellTEC 코칭 서비스 및 콘텐츠 개발
기 여 율 1/2
주관기관 순천향대학교
연구기간 2014.06.01 ~ 2018.12.31

이 발명을 지원한 국가연구개발사업
과제고유번호 20150004030041001
부처명 과학기술정보통신부
연구관리전문기관 정보통신기술진흥센터(IITP)
연구사업명 대학ICT연구센터육성 지원사업
연구과제명 IoT보안기술연구
기 여 율 1/2
주관기관 순천향대학교 산학협력단
연구기간 2015.06.01 ~ 2018.12.31

CNN and LSTM models for predicting particulate matter in the Republic of Korea

Title : CNN and LSTM models for predicting particulate matter in the Republic of Korea
Published in : The 8th International Conference on Information, System and Convergence Applications International Symposium on Innovation in Information Technology and Application 2020 (ICISCA 2020)
Author : Minsu Chae, Sangwook Han, HwaMin Lee
Corresponding author : HwaMin Lee
Location : Ton Duc Thang University, Ho Chi Minh, Vietnam and Online

Abstract
Recently, fine dust has become a problem all over the world. Many countries set regulations for PM10 and PM2.5. Because fine dust adversely affects human health, it is important to alert PM10 and PM2.5 in each city. Predictions for PM10 and PM2.5 are required to alert. We propose a hybrid model of the CNN(Convolutional Neural Network) model and the LSTM(Long Short-Term Memory) model to predict the PM10 and PM2.5 in the Republic of Korea. In addition, we evaluate the performance of the CNN model, the LSTM model, and the proposed hybrid model. As a result of the performance evaluation, the proposed hybrid model has high accuracy.

Acknowledgments
This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF-2017R1A2B4010570).

A performance comparison of linux containers and virtual machines using Docker and KVM

Title : A performance comparison of linux containers and virtual machines using Docker and KVM
Journal : Cluster Computing
Authors : MinSu Chae, HwaMin Lee, Kiyeol Lee
Corresponding author : HwaMin Lee
DOI : https://doi.org/10.1007/s10586-017-1511-2

Abstract
Virtualization is a foundational element of cloud computing. Since cloud computing is slower than a native system, this study analyzes ways to improve performance. We compared the performance of Docker and Kernel-based virtual machine (KVM). KVM uses full virtualization, including ×86 hardware virtualization extensions. Docker is a solution provided by isolation in userspace instead of creating a virtual machine. The performance of KVM and Docker was compared in three ways. These comparisons show that Docker is faster than KVM.

Acknowledgments
This work was supported by the National Research Foundation of Korea (NRF) Grant funded by the Korea government (Ministry of Science and ICT) (No. NRF-2017R1A2B4010570) and the Soonchunhyang University Research Fund.

Design of middleware to support auto-scaling in Docker-based multi host environment

Title : Design of middleware to support auto-scaling in Docker-based multi host environment
Published in : The 14th KIPS International Conference on Ubiquitous Information Technologies and Applications (CUTE 2019)
Author : Minsu Chae, Sangwook Han, HwaMin Lee
Corresponding author : HwaMin Lee
Location : University of Macau, Macau, China

Abstract
With the spread of smart devices, the use of big data, and the proliferation of the Internet of Things, virtualization technology for cloud servers have become important worldwide. Also, research has been conducted to efficiently manage the resources of hosts in VMs. Container-based virtualization has less performance degradation than VMs because there is no emulation for the operating system. Using the Docker API is slow to measure. In this paper, we implement the resource measurement module of Job nodes and design middleware that supports auto-scaling in auto-scaling module and Docker-based multi-host environment.

Acknowledgments
This research supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2019-2014-1-00720 & IITP-2019-2015-0-00403) supervised by the IITP(Institute for Information & communications Technology Planning & Evaluation).

A Design of Cardiac Arrest Early Warning System in Hospital

Title : A Design of Cardiac Arrest Early Warning System in Hospital
Published in : The 3rd International Conference on Interdisciplinary Research on Computer Science, Psychology, and Education (ICICPE’ 2019)
Author : SangWook Han, MinSu Chae, Min Hong, HwaMin Lee
Corresponding author : HwaMin Lee
Location : Vinoasis Phu Quoc Hotel, Phu Qoc, Vietnam

Abstract
Recently, increasing heart failure patients due to the increase in elderly patients, high probability for these patients, the arrhythmia. The number of cardiac arrests in hospitals is also increasing, and mortality rates are more than twice as high as those transferred from general wards to the intensive care unit through the emergency room or other routes. Existing patient monitoring methods in hospitals are measured using 24-hour Holter monitoring. This method requires carrying the device and measuring up to 48 hours. Also, the patient’s activities are very uncomfortable because of the communication line. Currently, many hospitals use the Modified Early Warning Score (MEWS) as a standard for early detection of deterioration in general ward patients. However, the prediction rate is very low and there is little help. Many people are working hard to detect heart attacks early because computerized systems vary from hospital to hospital, it is difficult to apply the developed system to various hospital environments. This study discusses the factors and methods necessary to solve these problems.

Acknowledgments
This research was supported by MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2019-2014-1-00720) supervised by the IITP(Institute for Information & communications Technology Planning & Evaluation) and the Bio & Medical Technology Development Program of the National Research Foundation (NRF)(No.NRF-2019M3E5D1A02069073).

CNN and RNN models for predicting particulate matter in South Korea

Title : CNN and RNN models for predicting particulate matter in South Korea
Published in : The 11th International Conference on Internet (ICONI 2019)
Author : Guang Yang, Thanongsak Xayasouk, HwaMin Lee
Corresponding author : HwaMin Lee
Location : Lotte Hotels & Resorts Hanoi, Hanoi, Vietnam

Abstract
Particulate matter proved to have severe effects on human health. To implement a real-time predicting system for is indispensable for every country face air pollution problem. Based on the previous study, predicting models for local areas are implemented. In this paper, a convolutional network, a recurrent neural network and a hybrid model combines the convolutional layers with the recurrent layers apply to predict future one weeks PM10 and PM2.5 concentration in seven urban cities in South Korea. The main contribution in this paper is: proposed three predicting models, make a comparison of three different models for predicting the concentration of particulate matter, all models are proved capable to obtain reliable predicting results. Experiments show GRU archive best performance for most cities both for PM10 and PM2.5, CNN models speed up the training process with least time consumption. The hybrid model performance stable with less time for training.

Acknowledgments
This research supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2019-2015-0-00403) supervised by the IITP(Institute for Information & communications Technology Planning & Evaluation) and Basic Science Research Program through the National Research Foundation of Korea (NRF-2017R1A2B4010570).