Digital Library[ Search Result ]
A Case Study of Industrial Software Defect Prediction in Maritime and Ocean Transportation Industries
Jonggu Kang, Duksan Ryu, Jongmoon Baik
http://doi.org/10.5626/JOK.2020.47.8.769
Software defect prediction is a field of study that predicts defects in newly developed software in advance of use, based on models trained with past software defects and software update information using various latest machine learning techniques. It can provide a guide to effectively operate and deploy software quality assurance (SQA) resources in industry practices. Recently, there have been papers that have investigated the industrial application of software defect prediction, but more active research is needed to analyze how this can be applied over diverse domains with different characteristics. In this paper, we present the possibility of applying software defect prediction in the maritime and ocean transportation industries. These are facing challenges to build and deploy the types of emerging transportations such as high-efficiency eco-friendly ships, connected ships, smart ships, unmanned ships, or autonomous ships. In our experiments using actual data collected from the domain, the software defect prediction showed high defect prediction performance with 0.91 accuracy and 0.831 f-measure. This suggests that software defect prediction can be a useful tool to allocate SQA resources effectively in this field.
Path Embedding-Based Knowledge Graph Completion Approach
Batselem Jagvaral, Min-Sung Kim, Young-Tack Park
http://doi.org/10.5626/JOK.2020.47.8.722
Knowledge graphs are widely used in question answering systems. However, in these circumstances most of the relations between the entities in the knowledge graph tend to be missing. To solve this issue, we propose a CNN(Convolutional Neural Network) + BiLSTM(Bidirectional LSTM) based approach to infer missing links in the knowledge graphs. Our method embeds paths connecting two entities into a low-dimensional space via CNN + BiLSTM. Then, an attention operation is used to attentively combine path embeddings to represent two entities. Finally, we measure the similarity between the target relation and representation of the entities to predict whether or not the relation connects those entities. By combining a CNN and BiLSTM, we are able to take advantage of the CNN’s ability to recognize local patterns and the LSTM’s ability to produce entity and relation ordering. In this way, it is possible to effectively identify low-dimensional path features and predict the relationships between entities using the learned features. In our experiments, we performed link prediction tasks on 4 different knowledge graphs and showed that our method achieves comparable results to state-of-the-art methods.
Approach for Managing Multiple Class Membership in Knowledge Graph Completion Using Bi-LSTM
Jae-Seung Roh, Batselem Jagvaral, Wan-Gon Lee, Young-Tack Park
http://doi.org/10.5626/JOK.2020.47.6.559
Knowledge graphs that represent real world information in a structured way are widely used in areas, such as Web browsing and recommendation systems. But there is a problem of missing links between entities in knowledge graphs. To resolve this issue, various studies using embedding techniques or deep learning have been proposed. Especially, the recent study combining CNN and Bidirectional-LSTM has shown high performance compared to previous studies. However, in the previous study, if multiple class types are defined for single entity, the amount of training data exponentially increases with the training time. Also, if class type information for an entity is not defined, training data for that entity cannot be generated. Thus, to enable the generation of training data for such entities and manage multiple class membership in knowledge graph completion, we propose two approaches using pre-trained embedding vectors of knowledge graph and the concept of vector addition. To evaluate the performance of the methods proposed in this paper, we conducted comparative experiments with the existing knowledge completion studies on NELL-995 and FB15K-237 datasets, and obtained MAP 1.6%p and MRR 1.5%p higher than that of the previous studies.
A Deep Learning LSTM Framework for Urban Traffic Flow and Fine Dust Prediction
Hongsuk Yi, Khac-Hoai Nam Bui, Choong-Nyoung Seon
http://doi.org/10.5626/JOK.2020.47.3.292
Accurate and timely forecasting is an essential step for the successful deployment of smart cities. With the rapid growth of traffic data collected daily, recent studies have focused on deep learning based on long-term short term memory (LSTM) for short-term traffic prediction, especially in urban areas. However, the short-term (five minutes) LSTM model is limited in the real-time nonlinear traffic flow prediction. Moreover, the fine dust prediction based on traffic data is also an emerging issue in this research area. Thus, this paper designs the multiple traffic data-based multi-input/output LSTM framework for supporting medium and long-term prediction. Additionally, a convolutional LSTM (ConvLSTM) model is developed for predicting fine dust flow based on traffic data. Regarding the experiment, we analyzed data from the Vehicle Detection System (VDS) located on major roads in Daejeon City for the evaluation. The experiment indicates promising results for the proposed approach.
Language Style Transfer Based on Surface-Level Neutralization
Wooyong Choi, Yunseok Noh, Seyoung Park
http://doi.org/10.5626/JOK.2020.47.2.162
Two main concerns of language style transfer such as sentiment transfer are neutralization of a stylized sentence and re-stylization of the neutralized sentence with a target style. Generally, neutralization is accomplished by learning a neutralized latent space by adversarial learning. However, this neutralization method suffers from the difficulty of maintaining the original content after style transfer. In this paper, we propose a two-step language style transfer method comprised of a surface-level neutralization that removes style words and a target-style word prediction for the removed words. For this, a self-attentive style classifier and style-specific word predictors are used for the surface neutralization and style word generation, respectively. To evaluate the proposed method, several experiments of language style transfer were conducted with Yelp and Amazon review datasets and Caption dataset. As a result, the proposed method shows superior performance over baseline methods on various evaluation metrics including automatic and human evaluations.
Latent Dirichlet Allocation Based Crime Code Clustering and Crime Prediction
http://doi.org/10.5626/JOK.2020.47.1.45
Predicting crime using crime data has become one of the most actively researched disciplines in major cites. Based on the research, law enforcement officials are shifting their efforts from crime investigation to crime prevention through predictive policing. Predictive policing highly relies on mathematics and statistics and identifies the underlying patterns of crimes. Based on these patterns, law enforcement officials can identify potential criminal activities. For accurate prediction, crime data must be well organized and managed. We first introduce one of the popular crime data set actively used by researchers. The data set categorizes each incident through a crime code. Examining the frequency of these codes allows regional agencies to predict the type of potential crimes, leading to effective predictive policing. In this research, we introduce a machine learning-based approach that can identify the similarity between the codes. Based on these similarities, we compute the frequencies of clusters and predict the code of potential crimes. Our experimental results show how our algorithm outperforms the statistical method.
Extraction of Cognitive Psychological Features of Mobile Gamers and Improvement of Purchases Prediction Performance
Jihoon Jeon, Seongil Yang, KyungJoong Kim
http://doi.org/10.5626/JOK.2019.46.9.892
In-game purchases are one of the important factors that directly affect a company"s revenue. In total, 95% of gamers do not pay for in-game purchases, meaning that a small number of gamers are responsible for most of the revenue of the company behind their games. For this reason, game companies must maintain and augment these few purchasing gamers. In this paper, we extracted seven cognitive psychological features (competitive, challenge, loyal, social, activity, efficient, and sincerity) that can be used to estimate the cognitive psychology of a gamer by using log data of a mobile RPG game. We analyzed the gamers, classified by payment amount, based on seven cognitive psychological features. As a result, the cognitive psychological features and payment amount of the gamers could be correlated. In addition, using seven cognitive psychological features, we predicted the purchasing behavior of gamers with high accuracy. This implies that gamers can be analyzed based on their cognitive psychology and the gamer"s purchases can be predicted with comparatively high performance.
Design of Photovoltaic Power Generation Prediction Model with Recurrent Neural Network
Hanho Kim, Haesung Tak, Hwan-gue Cho
http://doi.org/10.5626/JOK.2019.46.6.506
The Smart Grid predicts the power generation amount of renewable energy and enables efficient power generation and consumption. Existing PV power generation prediction studies have rarely applied and compared recurrent neural network techniques that are superior to time series. Furthermore, in the reported studies, there is no consideration of the length of past data used for learning, leading to lowered prediction performance of the model. In this study, we used the embedded variable selection techniques to find the factors influencing PV power generation. Subsequently, experiments were carried out to insert various past data length into the recurrent neural networks (RNN, LSTM, GRU). We found the optimal prediction factors and designed a prediction model based on the outcomes of the experiments. The designed PV power generation prediction model shows better prediction performance compared to other factor settings. In addition, better performance based on the prediction rate is confirmed in the present study as compared with the existing researches.
User Behavior Analysis for Predicting Churn of Loyal Customers in Online Games based on Social Relationships and Degree of Participation
Eunbi Seo, Jiyoung Woo, Huy Kang Kim
http://doi.org/10.5626/JOK.2018.45.11.1124
Game users in MMORPGs engage in a variety of social activities. However, a few users tend to play games alone, and are designated ‘loners’ similar to modern society. We classified game guilds and game users based on similar user behaviors and community characteristics. We propose a model that predicts churn users by measuring the participation of users in each group. Users in each group show similar behavioral patterns, suggesting that we can classify churn users along with ordinary users. We tested this model for NCsoft’s MMORPG, Aion. Using Randomforest, the recall was measured at an average of 75%.
Automatic Segmentation of Renal Parenchyma using Shape and Intensity Information based on Multi-atlas in Abdominal CT Images
Hyeonjin Kim, Helen Hong, Kidon Chang, Koon Ho Rha
http://doi.org/10.5626/JOK.2018.45.9.937
Renal parenchyma segmentation is necessary to predict contralateral hypertrophy after renal partial nephrectomy. In this paper, we propose an automatic segmentation method of renal parenchyma using shape and intensity information based on the multi-atlas in abdominal CT images. First, similar atlases are selected using volume-based similarity registration and intensity-similarity measure. Second, renal parenchyma is segmented using two-stage registration and constrained intensity-based locally-weighted voting. Finally, renal parenchyma is refined using a Gaussian mixture model-based multi-thresholds and shape-prediction map in under- and over-segmented data. The average dice similarity coefficient of renal parenchyma was 91.34%, which was 18.19%, 1.35% higher than the segmentation method using majority voting and locally-weighted voting in dice similarity coefficient, respectively.
Search
Journal of KIISE
- ISSN : 2383-630X(Print)
- ISSN : 2383-6296(Electronic)
- KCI Accredited Journal
Editorial Office
- Tel. +82-2-588-9240
- Fax. +82-2-521-1352
- E-mail. chwoo@kiise.or.kr