No abstract

As people communicate with each other, they use gestures and facial expressions as a means to convey and understand emotional state. Non-verbal means of communication are essential to understanding, based on external clues to a person’s emotional state. Recently, active studies have been conducted on the lifecare service of analyzing users’ facial expressions. Yet, rather than a service necessary for everyday life, the service is currently provided only for health care centers or certain medical institutions. It is necessary to conduct studies to prevent accidents that suddenly occur in everyday life and to cope with emergencies. Thus, we propose facial expression analysis using line-segment feature analysis-convolutional recurrent neural network (LFA-CRNN) feature extraction for health-risk assessments of drivers. The purpose of such an analysis is to manage and monitor patients with chronic diseases who are rapidly increasing in number. To prevent automobile accidents and to respond to emergency situations due to acute diseases, we propose a service that monitors a driver’s facial expressions to assess health risks and alert the driver to risk-related matters while driving. To identify health risks, deep learning technology is used to recognize expressions of pain and to determine if a person is in pain while driving. Since the amount of input-image data is large, analyzing facial expressions accurately is difficult for a process with limited resources while providing the service on a real-time basis. Accordingly, a line-segment feature analysis algorithm is proposed to reduce the amount of data, and the LFA-CRNN model was designed for this purpose. Through this model, the severity of a driver’s pain is classified into one of nine types. The LFA-CRNN model consists of one convolution layer that is reshaped and delivered into two bidirectional gated recurrent unit layers. Finally, biometric data are classified through softmax. In addition, to evaluate the performance of LFA-CRNN, the performance was compared through the CRNN and AlexNet Models based on the University of Northern British Columbia and McMaster University (UNBC-McMaster) database.

In order to cope with disaster situations properly, it is very important to identify the disaster scale and provide the accurate information of the site to the appropriate authorities including disaster site and Central Disaster Management Center, on-site command post, etc. and share the information provided. In particular, sharing information on disaster situations should control the disaster quickly to prevent the disaster situation from lasting and expanding. However, in the event of a large-scale disaster, delay is caused in the existing commercial network and therefore, the disaster situation cannot be communicated quickly and accurately. In order to determine the situation exactly in the event of a disaster, safety and connectivity of the network and flow of data are very important. Even if the stability of the network and connection of nodes are resolved in the network of each agency business operator, it is necessary to share the platform between networks for IoT/ M2M communication for the smooth flow of data. Recently, the disaster safety net of combining existing disaster standard technology with Ubiquitous Technology and Smart IT such as Tetra of Europe, iDEN of the U.S., etc. has been built for disaster safety communications. In addition, systems useful for demand-centered, site-centered immediate disaster response by using Mobile, SNS, cloud computing, etc. are being built and designed to play an important role in the disaster information system especially through IoT, P2P cloud network, big data, etc. Therefore, in this paper, we proposed the P2P cloud network service for IoT based disaster situations information according to the paradigm of the changing times. The proposed service is to combine IoT/M2M network with P2P cloud service for rapid and smooth response in the event of a disaster and provide the results as social services such as SNS. To this end, the wide area wireless disaster information network system has been built in the local and each local network is connected to each other to provide disaster situations by using the server of the disaster area. At this time, each server was to be interconnected via P2P network and to be connected automatically by software-based network in P2P Cloud System. Also, the cognitive cycle was applied for selecting optimal wireless link and router of P2P Cloudbased Disaster Information Network and the danger situations of the disaster area were to be provided to the user by configuring disaster information component for providing services and building central disaster information platform managing it.

With the rapid increase in vehicle use during the fourth Industrial Revolution, road resources have reached their supply limit. Active studies have therefore been conducted on intelligent transportation systems (ITSs) to realize traffic management systems utilizing fewer resources. As part of an ITS, real-time traffic services are provided to improve user convenience. Such services are applied to prevent traffic congestion and disperse existing traffic. Therefore, these services focus on immediacy at the expense of accuracy. As these services typically rely on measured data, the accuracy of the models are contingent on the data collection. Therefore, this study proposes a long short-term memory (LSTM)-based traffic congestion prediction approach based on the correction of missing temporal and spatial values. Before making predictions, the proposed prediction method applies pre-processing that consists of outlier removal using the median absolute deviation of the traffic data and the correction of temporal and spatial values using temporal and spatial trends and pattern data. In previous studies, data with time-series features have not been appropriately learned. To address this problem, the proposed prediction method uses an LSTM model for time-series data learning. To evaluate the performance of the proposed method, the mean absolute percentage error (MAPE) was calculated for comparison with other models. The MAPE of the proposed method was found to be the best of the compared models, at approximately 5%.

Arrhythmia detection through deep learning is mainly classified through supervised learning. Supervised learning progresses through the labeled data. However, in the medical field, it is challenging to collect ECG data of patients with arrhythmia than ECG data of healthy people, and thus data bias occurs. Therefore, if you use a supervised learning model, there are problems with lack of data and imbalance between labels that arise during learning. Accordingly, this study proposes the decision boundary-based Anomaly detection model using improved AnoGan from ECG data. In this study, at the time of learning, the loss of the Generator does not reduce, but the loss of a Discriminator lowers. Even if the Generator and Discriminator were designed to have the same learning count, the learning competency of Generator was judged to be lowered. In repeated experiments, it was found that the best loss balance was achieved when the learning count of Discriminator was 1 and that of Generator was 4. Another problem is that the decision boundary of AnoGAN is subjective. Accordingly, the repeated experiments based on F-measure are conducted to determine a decision boundary. For performance evaluation, the accuracy of the model is evaluated on the basis of Epoch, and the goodness-of-fit of the model is evaluated on the basis of AUC and F-measure. According to the evaluation of F-measure, the model has the best performance when the decision boundary is 200. In terms of Epoch, the model has the highest accuracy when the Epoch is 10. In addition, the proposed model has better goodness-of-fit than AnoGAN.