Recently, people with disabilities also search for and collect information from the web through smart devices, and save collected information on smart devices or take notes. For non-disabled people, various memo applications are provided on the market, so it is more convenient to choose according to their preference. However, existing memo services are limited for use by blind people due to the importance of visual information. The problem with blind people when using smart devices is that the screen is not recognized, so it is not possible to check in which location the menu of the application exists. In addition, it is difficult to input and manipulate text, and systematic file management and control are not possible. Therefore, in this paper, we propose the development of voice memo service that blind people can use only voice and hearing information and can operate menu with Bluetooth remote controller. We will develop a system that includes a comprehensive voice file management function for storing, searching, playing, and deleting files, rather than simply storing voice files.

The injection is the input method of the image feature vector from the encoder to the decoder. Since the image feature vector contains object details such as color and texture, it is essential to generate image captions. However, the bidirectional decoder model using the existing injection method only inputs the image feature vector in the first step, so image feature vectors of the backward sequence are vanishing. This problem makes it difficult to describe the context in detail. Therefore, in this paper, we propose the parallel injection method to improve the description performance of image captions. The proposed Injection method fuses all embeddings and image vectors to preserve the context. Also, We optimize our image caption model with Bidirectional Gated Recurrent Unit (Bi-GRU) to reduce the amount of computation of the decoder. To validate the proposed model, experiments were conducted with a certified image caption dataset, demonstrating excellence in comparison with the latest models using BLEU and METEOR scores. The proposed model improved the BLEU score up to 20.2 points and the METEOR score up to 3.65 points compared to the existing caption model.

The goal of the sound-based mechanical fault diagnosis technique is to automatically find abnormal signals in the machine using acoustic emission. Conventional methods of using mathematical models have been found to be inaccurate due to the complexity of industrial mechanical systems and the existence of nonlinear factors such as noise. Therefore, any fault diagnosis issue can be treated as a pattern recognition problem. We propose an automatic fault diagnosis method using discrete wavelet transform and power spectrum density using multi pattern recognition. First, we perform DWT-based filtering analysis for noise cancelling and effective feature extraction. Next, the power spectral density(PSD) is performed on each subband of the DWT in order to effectively extract feature vectors of sound. Finally, each PSD data is extracted with the features of the classifier using multi pattern recognition. The results show that the proposed method can not only be used effectively to detect faults as well as apply to various automatic diagnosis system based on sound.

Super resolution technique aims to convert a low-resolution image with coarse details to a corresponding high-resolution image with refined details. In the past decades, the performance is greatly improved due to progress of deep learning models. However, universal solution for various objects is a still challenging issue. We observe that learning super resolution with a general dataset has poor performance on faces. In this paper, we propose a super resolution fusion scheme that works well for both general- and face datasets to achieve more universal solution. In addition, object-specific feature extractor is employed for better reconstruction performance. In our experiments, we compare our fusion image and super-resolved images from one- of the state-of-the-art deep learning models trained with DIV2K and FFHQ datasets. Quantitative and qualitative evaluates show that our fusion scheme successfully works well for both datasets. We expect our fusion scheme to be effective on other objects with poor performance and this will lead to universal solutions.

Intelligent tutoring system enables users to effectively learn by utilizing various artificial intelligence techniques. For instance, it can recommend a proper curriculum or learning method to individual users based on their learning history. To do this effectively, user's characteristics need to be analyzed and classified based on various aspects such as interest, learning ability, and personality. Even though data labeled by the characteristics are required for more accurate classification, it is not easy to acquire enough amount of labeled data due to the labeling cost. On the other hand, unlabeled data should not need labeling process to make a large number of unlabeled data be collected and utilized. In this paper, we propose a semi-supervised learning method based on feedback variational auto-encoder(FVAE), which uses both labeled data and unlabeled data. FVAE is a variation of variational auto-encoder(VAE), where a multi-layer perceptron is added for giving feedback. Using unlabeled data, we train FVAE and fetch the encoder of FVAE. And then, we extract features from labeled data by using the encoder and train classifiers with the extracted features. In the experiments, we proved that FVAE-based semi-supervised learning was superior to VAE-based method in terms with accuracy and F1 score.

Recently, Communication media have been developed for the underwater communication in underwater environments where the underwater communication failures occur frequently. Although the underwater communication by one medium is not stable due to the influence of the underwater environment, the use of various communication media can complement each other and so this makes the underwater communication more stable. For this reason, this paper proposes approaches to the design and implementation of integrated MAC for complementing individual unstable underwater communications. In addition this paper presents the comparison and analysis on each of the proposed approaches so to be able to provide guidelines to designers and implementers of integrated MAC.

In medical image analysis, segmentation is considered as a vital process since it partitions an image into coherent parts and extracts interesting objects from the image. In this paper, we consider automatic segmentations of OCT retinal images to find six layer boundaries using convolutional neural networks. Segmenting retinal images by layer boundaries is very important in diagnosing and predicting progress of eye diseases including diabetic retinopathy, glaucoma, and AMD (age-related macular degeneration). We applied well-known CNN architecture for general image segmentation, called Segnet, U-net, and CNN-S into this problem. We also proposed a shortest path-based algorithm for finding the layer boundaries from the outputs of Segnet and U-net. We analysed their performance on public OCT image data set. The experimental results show that the Segnet combined with the proposed shortest path-based boundary finding algorithm outperforms other two networks.

For developing a wireless implantable device to monitoring the artery variation in real-time. The concept of a special vessel variation measurement capacitive sensor is presented in this paper. The sensor consists of two part; main sensor to measuring the arterial variation, and reference sensor is used to improve the accuracy of the capacitance value variation. Before sensor manufacture, a model of the sensor attached on the artery was designed in 3D to conduct in the FEA simulation to validate the validity and feasibility of the idea. The artery model was designed as layered structures and made of collagenous soft tissues with intima inside, followed by the media and the adventitia. Also, a grease layer was designed in the inner of the arterial wall to imitate the clogged arteries. The simulation was divided into two parts; sensor performance test by changing the diameter of the grease layer, and arterial wall tension test by changing the blood pressure. As the simulation results, the capacitance value measured by the proposed sensor is decreased follow the diameter of the grease increased. Also, large elastic deformation of the arterial wall since changing the blood pressure has been observed.

As life expectancy increases, chronic diseases become a problem, and health management in daily life becomes important. With the development of IT convergence technology, personal health records have been collected through various data sources, but they are saved separately by each medical services. Distributed storage of personal health records caused inconvenience about managing the user's health records. In this paper, we designed and implemented an integrated repository and web-based dashboard UI components to solve that inconvenience. The proposed method shows that users can manage their personal health records in the integrated repository effectively and show them through dashboard UI components.

BSTRACT In this paper, PM10 forecast model using DNN(Deep Neural Network) is developed for Seoul region. The previous Julian forecast model has been developed using weather and air quality data of Seoul region only. This model gives excellent results for accuracy and false alarm rates, but poor result for POD(Probability of Detection). To solve this problem, an WA(Wide Area) forecasting model that uses Chinese data is developed. The data is highly correlated with the emergence of high concentrations of PM10 in Korea. As a result, the WA model shows better accuracy, and POD improving of 3%(D+0), 21%(D+1), and 36%(D+2) for each forecast period compared with the Julian model.

With the development of the Internet and the increase of smart phones, various services considering user convenience are increasing, so that users can check news in real time anytime and anywhere. However, online news is categorized by media and category, and it provides only a few related search terms, making it difficult to find related news related to keywords. In order to solve this problem, we propose a method to recommend related documents more accurately by applying Doc2Vec similarity to the specific keywords of news articles and weighting the title and contents of news articles. We collect news articles from Naver politics category by web crawling in Java environment, preprocess them, extract topics using LDA modeling, and find similarities using Doc2Vec. To supplement Doc2Vec, we apply TF-IDF to obtain TC(Title Contents) weights for the title and contents of news articles. Then we combine Doc2Vec similarity and TC weight to generate TC weight-similarity and evaluate the similarity between words using PMI technique to confirm the keyword association.

This study set out to investigate an energy harvesting device to ensure stable energy supply to batteries and data collection devices and a monitoring system for acer pictum sap to check collected data. Acer pictum sap farmers have written down weather information and yield of acer pictum sap manually for data storage. Since the job is done manually, there are many missing values in their data. In addition, it is not easy to manage batteries due to the characteristics of the areas where acer pictum sap is collected. The present study thus decided to build an energy harvesting device based on new renewable energy to ensure stable energy supply by taking into consideration power load, daily power consumption, and number of days with no sunshine for various devices. For a monitoring system, the investigator proposed a JSP-based web page to monitor temperature, humidity, volume of collected water, and battery state in real time. The proposed energy harvesting device was applied to reduce missing values in data. It promoted stable energy supply to the batteries and data collection devices, reducing the percentage of missing values in data from 30.55% to 0%.