• Journal of the Korea Society for Simulation
• /
• v.32 no.3
• /
• pp.43-53
• /
• 2023

In this detailed and comprehensive study, our primary focus has been placed on accurately gauging the number of visitors and their real-time locations in commercial spaces. Particularly, in a real cafe, using security cameras, we have developed a system that can offer live updates on available seating and predict future congestion levels. By employing YOLO, a real-time object detection and tracking algorithm, the number of visitors and their respective locations in real-time are also monitored. This information is then used to update a cafe's indoor map, thereby enabling users to easily identify available seating. Moreover, we developed a model that predicts the congestion of a cafe in real time. The sophisticated model, designed to learn visitor count and movement patterns over diverse time intervals, is based on Long Short Term Memory (LSTM) to address the vanishing gradient problem and Sequence-to-Sequence (Seq2Seq) for processing data with temporal relationships. This innovative system has the potential to significantly improve cafe management efficiency and customer satisfaction by delivering reliable predictions of cafe congestion to all users. Our groundbreaking research not only demonstrates the effectiveness and utility of indoor location tracking technology implemented through security cameras but also proposes potential applications in other commercial spaces.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.05a
• /
• pp.431-432
• /
• 2018

MRC(기계독해)는 Passage, Question, Answel 로 이루어진 Dataset 으로 학습된 모델을 사용하여 요청한 Question 의 Answer 를 같이 주어진 Passage 내에서 찾아내는 것을 목적으로 한다. 최근 MRC 시스템의 성능 측정 지표로 활용되는 SQuAD Dataset 을 활용하여 RNN 의 한 분류인 match-LSTM과 R-NET 알고리즘의 성능을 비교 분석하고자 한다.

• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.2
• /
• pp.45-52
• /
• 2020

Since the opening of the national petition site, it has attracted much attention. In this paper, we perform topic analysis of the national petition site and propose a prediction model for answerable petitions based on deep learning. First, 1,500 petitions are collected, topics are extracted based on the petitions' contents. Main subjects are defined using K-means clustering algorithm, and detailed subjects are defined using topic modeling of petitions belonging to the main subjects. Also, long short-term memory (LSTM) is used for prediction of answerable petitions. Not only title and contents but also categories, length of text, and ratio of part of speech such as noun, adjective, adverb, verb are also used for the proposed model. Our experimental results show that the type 2 model using other features such as ratio of part of speech, length of text, and categories outperforms the type 1 model without other features.

• Journal of Broadcast Engineering
• /
• v.24 no.6
• /
• pp.1013-1023
• /
• 2019

In content consumption environment with various live TV channels, VoD contents and web contents, recommendation service is now a necessity, not an option. Currently, various kinds of recommendation services are provided in the OTT service or the IPTV service, such as recommending popular contents or recommending related contents which similar to the content watched by the user. However, in the case of a content viewing environment through TV or IPTV which shares one TV and a TV set-top box, it is difficult to recommend proper content to a specific user because one or more usage histories are accumulated in one subscription information. To solve this problem, this paper interprets the concept of family as {user, time}, extends the existing recommendation relationship defined as {user, content} to {user, time, content} and proposes a method based on deep learning algorithm. Through the proposed method, we evaluate the recommendation performance qualitatively and quantitatively, and verify that our proposed model is improved in recommendation accuracy compared with the conventional method.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.5
• /
• pp.187-194
• /
• 2019

Traditional demand forecasting methods are difficult to meet the needs of companies due to rapid changes in the market and the diversification of individual consumer needs. In a diversified production environment, the right demand forecast is an important factor for smooth yield management. Many of the existing predictive models commonly used in industry today are limited in function by little. The proposed model is designed to overcome these limitations, taking into account the part where each model performs better individually. In this paper, variables are extracted through Gray Relational analysis suitable for dynamic process analysis, and statistically predicted data is generated that includes characteristics of historical demand data produced through ARIMA forecasts. In combination with the LSTM model, demand forecasts can then be calculated by reflecting the many factors that affect demand forecast through an architecture that is structured to avoid the long-term dependency problems that the neural network model has.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2019.05a
• /
• pp.39-39
• /
• 2019

인구의 도시 집중화로 인하여 다량의 생활용수의 사용에 따라 하천의 자정능력을 초과하여 오염을 유발시키고 있다. 이에 도시하천들의 오염은 점점 심해져 경제적으로 많은 문제를 유발하고 있다. 이러한 하천오염 문제를 과학적으로 대응하기 위해서는 오염물질의 농도 측정 및 데이터 축척을 통한 오염예측이 필수적이라 할 수 있으며, 부산광역시 보건환경정보 공개시스템에서는 하천수질 자동측정망을 설치하여 시간 단위로 오염물질을 측정하고 있다. 그러나 온천천의 하천수질 데이터는 계속 쌓여가고 있는데 이 데이터를 활용해서 하천수질 인자 예측이 거의 이뤄지지 않고 있다. 본 연구에서는 순환신경망 알고리즘을 활용하여 일 단위의 하천수질 인자 예측을 시도하였다. 순환신경망은 인공신경망의 발전된 형태인 시계열 학습에 강한 RNN, LSTM 알고리즘을 활용한 일단위 하천수질 인자 예측을 하고자 하였다. 연구에 앞서 시간 단위로 쌓여있는 데이터를 평균 내어 일 단위로 변경하였고 이 데이터를 가지고 일 단위 하천수질 인자 예측을 진행하였다. 연구에는 Google에서 개발한 딥러닝 오픈소스 라이브러리인 텐서플로우를 활용하여 DO, 탁도 등 항목을 예측하였다. 하천오염의 학습과 예측을 위해 대상지로는 부산지역 온천천의 부곡교, 세병교, 이섭교 관측소를 선택하였다. 연구를 위해 DO, 탁도 등 자료 수집은 부산광역시 보건환경정보 공개시스템의 자료를 활용하였다. 모형의 학습을 위해 입력자료로는 하천수질 인자 자료를 이용하였고, 자료의 학습에는 2014년~2017년 4년간의 자료를 학습자료로 사용하였고, 2018년 1년간의 자료는 모형의 검증을 위해 사용하였다. RNN, LSTM 알고리즘을 활용하여 분석 시 은닉층의 개수, 반복시행횟수, sequence length 등의 값을 조절하여 하천수질 인자 예측을 하였다. 모형의 검증을 위해 $R^2$(r square)와 RMSE(root mean square error)을 이용하여 통계분석을 실시하였다.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.4
• /
• pp.19-27
• /
• 2020

In this paper, we propose a machine learning method for diagnosing the failure of a gas pressure regulator. Originally, when implementing a machine learning model for detecting abnormal operation of a facility, it is common to install sensors to collect data. However, failure of a gas pressure regulator can lead to fatal safety problems, so that installing an additional sensor on a gas pressure regulator is not simple. In this paper, we propose various machine learning approach for diagnosing the abnormal operation of a gas pressure regulator with only the flow rate and gas pressure data collected from a gas pressure regulator itself. Since the fault data of a gas pressure regulator is not enough, the model is trained in all classes by applying the over-sampling method. The classification model was implemented using Gradient boosting, 1D Convolutional Neural Networks, and LSTM algorithm, and gradient boosting model showed the best performance among classification models with 99.975% accuracy.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.7
• /
• pp.869-876
• /
• 2021

In the field of music, various AI composition methods using machine learning have recently been attempted. However, most of this research has been centered on Western music, and little research has been done on Korean traditional music. Therefore, in this paper, we will create a data set of Korean traditional music, create a melody using three algorithms based on the data set, and compare the results. Three models were selected based on the similarity between language and music, LSTM, Music Transformer and Self Attention. Using each of the three models, a melody generator was modeled and trained to generate melodies. As a result of user evaluation, the Self Attention method showed higher preference than the other methods. Data set is very important in AI composition. For this, a Korean traditional music data set was created, and AI composition was attempted with various algorithms, and this is expected to be helpful in future research on AI composition for Korean traditional music.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.413-415
• /
• 2022

Since human joints can be used as useful information for analyzing human behavior as a component of the human body, many studies have been conducted on human action recognition using joint information. However, it is a very complex problem to recognize human action that changes every moment using only each independent joint information. Therefore, an additional information extraction method to be used for learning and an algorithm that considers the current state based on the past state are needed. In this paper, we propose a human action recognition technique considering the positional relationship of connected joints and the change of the position of each joint over time. Using the pre-trained joint extraction model, position information of each joint is obtained, and bone information is extracted using the difference vector between the connected joints. In addition, a simplified neural network is constructed according to the two types of inputs, and spatio-temporal features are extracted by adding LSTM. As a result of the experiment using a dataset consisting of 9 behaviors, it was confirmed that when the action recognition accuracy was measured considering the temporal and spatial relationship features of each joint, it showed superior performance compared to the result using only single joint information.

• Structural Monitoring and Maintenance
• /
• v.5 no.1
• /
• pp.51-65
• /
• 2018

A number of sensing techniques have been implemented for detecting defects in civil infrastructures instead of onsite human inspections in structural health monitoring. However, the issue of faults in sensors has not received much attention. This issue may lead to incorrect interpretation of data and false alarms. To overcome these challenges, this article presents a deep learning-based method with a new architecture of Stateful Long Short Term Memory Neural Networks (S-LSTM NN) for detecting sensor fault without going into details of the fault features. As LSTMs are capable of learning data features automatically, and the proposed method works without an accurate mathematical model. The detection of four types of sensor faults are studied in this paper. Non-stationary acceleration responses of a three-span continuous bridge when under operational conditions are studied. A deep network model is applied to the measured bridge data with estimation to detect the sensor fault. Another set of sensor output data is used to supervise the network parameters and backpropagation algorithm to fine tune the parameters to establish a deep self-coding network model. The response residuals between the true value and the predicted value of the deep S-LSTM network was statistically analyzed to determine the fault threshold of sensor. Experimental study with a cable-stayed bridge further indicated that the proposed method is robust in the detection of the sensor fault.