• Journal of KIISE:Software and Applications
• /
• v.35 no.3
• /
• pp.179-188
• /
• 2008

An important issue in multiagent reinforcement learning is how an agent should team its optimal policy through trial-and-error interactions in a dynamic environment where there exist other agents able to influence its own performance. Most previous works for multiagent reinforcement teaming tend to apply single-agent reinforcement learning techniques without any extensions or are based upon some unrealistic assumptions even though they build and use explicit models of other agents. In this paper, basic concepts that constitute the common foundation of multiagent reinforcement learning techniques are first formulated, and then, based on these concepts, previous works are compared in terms of characteristics and limitations. After that, a policy model of the opponent agent and a new multiagent reinforcement learning method using this model are introduced. Unlike previous works, the proposed multiagent reinforcement learning method utilize a policy model instead of the Q function model of the opponent agent. Moreover, this learning method can improve learning efficiency by using a simpler one than other richer but time-consuming policy models such as Finite State Machines(FSM) and Markov chains. In this paper. the Cat and Mouse game is introduced as an adversarial multiagent environment. And effectiveness of the proposed multiagent reinforcement learning method is analyzed through experiments using this game as testbed.

• Journal of Distribution Science
• /
• v.13 no.11
• /
• pp.123-130
• /
• 2015

Purpose - Due to highly elevated levels of competition, many companies today have to face the problem of decreasing profits even when their actual sales volume is increasing. This is a common phenomenon that is seen occurring among companies that focus heavily on quantitative growth rather than qualitative growth. These two aspects of growth should be well balanced for a company to create a sustainable business model. For supply chain management (SCM) planners, the optimized, quantified flow of resources used to be of major interest for decades. However, this trend is rapidly changing so that managers can put the appropriate balance between sales volume and sales quality, which can be evaluated from the profit margin. Profit optimization is a methodology for companies to use to achieve solutions focused more on profitability than sales volume. In this study, we attempt to provide executional insight for companies considering implementation of the profit optimization system to enhance their business profitability. Research design, data, and methodology - In this study, we present a comprehensive explanation of the subject of profit optimization, including the fundamental concepts, the most common profit optimization logic algorithm -linear programming -the business functional scope of the profit optimization system, major key success factors for implementing the profit optimization system at a business organization, and weekly level detailed business processes to actively manage effective system performance in achieving the goals of the system. Additionally, for the purpose of providing more realistic and practical information, we carefully investigate a profit optimization system implementation case study project fulfilled for company S. The project duration was about eight months, with four full-time system development consultants deployed for the period. To guarantee the project's success, the organization adopted a proven system implementation methodology, supply chain management (SCM) six-sigma. SCM six-sigma was originally developed by a group of talented consultants within Samsung SDS through focused efforts and investment in synthesizing SCM and six-sigma to improve and innovate their SCM operations across the entire Samsung Organization. Results - Profit optimization can enable a company to create sales and production plans focused on more profitable products and customers, resulting in sustainable growth. In this study, we explain the concept of profit optimization and prerequisites for successful implementation of the system. Furthermore, the efficient way of system security administration, one of the hottest topics today, is also addressed. Conclusion - This case study can benefit numerous companies that are eagerly searching for ways to break-through current profitability levels. We cannot guarantee that the decision to deploy the profit optimization system will bring success, but we can guarantee that with the help of our study, companies trying to implement profit optimization systems can minimize various possible risks across various system implementation phases. The actual system implementation case of the profit optimization project at company S introduced here can provide valuable lessons for both business organizations and research communities.

• The Journal of the Korea Contents Association
• /
• v.7 no.10
• /
• pp.311-320
• /
• 2007

This study is to analyze the present condition of internal dance troupes' web sites contents and through this study, we aim at suggesting beneficial plans not only for constructing better web sites and contents for national dance troupes but also for contributing to the development of national dance troupes. In order to achieve this study's object, we selected eight dance troupes that run their own web sites and examined their contents. the conclusions are as following. First, among internal professional dance troupes, Korea National Ballet, Seoul Ballet Theatre and Universal Ballet Company offer the best contents at their home pages and they have the most lively community concerned activities at their internet web sites. Second as for the web sites of municipal(provincial) dance troupes, those of Gyeonggi provincial dance troupe, Daejeon municipal dance troupe, Busan municipal dance troupe, Seoul municipal dance troupe Incheon municipal dance troupe have lots of contents briskly carried on, but those of the others need complementary measures. Third, Seoul ballet theatre and Universal Ballet Company offer lots of information about their activities to satisfy their customers and make the maximum use for public relations ; performance news, the introduction of their members, the story of ballet press reviews, new letters. On the other hand, at the web sites of Metropolitan Dance Theater Provincial Dance Company, there are much fewer contents about their activities which doesn't lead customers to visit their webs.

• Journal of Korean Society of Environmental Engineers
• /
• v.32 no.7
• /
• pp.690-698
• /
• 2010

Contamination of groundwater resources by organic chemicals has become an issue of increasing environmental concern. Surfactant-enhanced aquifer remediation (SEAR) is widely recognized as one of the most promising techniques to remediate organic contaminations in-situ. Solutions of surfactant or surfactant with polymer are used to dramatically expedite the process, which in turn, may reduce the treatment time of a site compared to use of water alone. In the design of surfactant-based technologies for remediation of organic contaminated aquifers, it is very important to have a considerable analysis using extensive numerical simulations prior to full-scale implementation. This study investigated the formation and flow of microemulsions during SEAR of organic-contaminated aquifer using the finite difference model UTCHEM, a three-dimensional, multicomponent, multiphase, compositional model. The remediation process variables considered in this study were the sequence of injection fluids, the injection and extraction rate, the concentrations of polymer in surfactant slug and chase water, and the duration of surfactant injection. For each variable, temporal changes in injection and production wells and spatial distributions of relative saturations in the organic phase were compared. Cleanup time and cumulative organic recovery were also quantified. The study would provide useful information to design strategies for the remediation of nonaqueous phase liquid-contaminated aquifers.

• Journal of KIISE:Software and Applications
• /
• v.35 no.11
• /
• pp.671-680
• /
• 2008

Motion capture devices have been utilized in producing several contents, such as movies and video games. However, since motion capture devices are expensive and inconvenient to use, motions segmented from captured data was recycled and synthesized to utilize it in another contents, but the motions were generally segmented by contents producers in manual. Therefore, automatic motion segmentation is recently getting a lot of attentions. Previous approaches are divided into on-line and off-line, where ow line approaches segment motions based on similarities between neighboring frames and off-line approaches segment motions by capturing the global characteristics in feature space. In this paper, we propose a graph-based high-level motion segmentation method. Since high-level motions consist of repeated frames within temporal distances, we consider similarities between neighboring frames as well as all similarities among all frames within the temporal distance. This is achieved by constructing a graph, where each vertex represents a frame and the edges between the frames are weighted by their similarity. Then, normalized cuts algorithm is used to partition the constructed graph into several sub-graphs by globally finding minimum cuts. In the experiments, the results using the proposed method showed better performance than PCA-based method in on-line and GMM-based method in off-line, as the proposed method globally segment motions from the graph constructed based similarities between neighboring frames as well as similarities among all frames within temporal distances.

• Journal of KIISE:Software and Applications
• /
• v.36 no.4
• /
• pp.253-260
• /
• 2009

Many algorithms for computer vision and pattern recognition have recently been implemented on GPU (graphic processing unit) for faster computational times. However, the implementation has two problems. First, the programmer should master the fundamentals of the graphics shading languages that require the prior knowledge on computer graphics. Second, in a job that needs much cooperation between CPU and GPU, which is usual in image processing and pattern recognition contrary to the graphic area, CPU should generate raw feature data for GPU processing as much as possible to effectively utilize GPU performance. This paper proposes more quick and efficient implementation of neural networks on both GPU and multi-core CPU. We use CUDA (compute unified device architecture) that can be easily programmed due to its simple C language-like style instead of GPU to solve the first problem. Moreover, OpenMP (Open Multi-Processing) is used to concurrently process multiple data with single instruction on multi-core CPU, which results in effectively utilizing the memories of GPU. In the experiments, we implemented neural networks-based text extraction system using the proposed architecture, and the computational times showed about 15 times faster than implementation on only GPU without OpenMP.

• KIPS Transactions on Software and Data Engineering
• /
• v.5 no.4
• /
• pp.171-180
• /
• 2016

As social networking services such as twitter become increasingly popular, data streams are widely prevalent these days. In order to search data accumulated from data streams efficiently, the use of an index structure is essential. In this paper, we propose an update-efficient, disk-based inverted index structure for efficient keyword search on data streams. When new data arrive at the data stream, the index needs to be updated to incorporate the new data. The traditional inverted index is very inefficient to update in terms of disk I/O, because all index data stored in the disk need to be read and written to the disk each time the index is updated. To solve this problem, we divide the whole inverted index into a sequence of inverted indices with exponentially increasing size. When new data arrives, it is first inserted into the smallest index and, later, the small indices are merged with the larger indices, which leads to a small amortize update cost for each new data. Furthermore, when indices stored in the disk are merged with each other, we minimize the disk I/O cost incurred for the merge operation, resulting in an even smaller update cost. Through various experiments, we compare the update efficiency of the proposed index structure with the previous one, and show the performance advantage of the proposed structure in terms of the update cost.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.10
• /
• pp.387-396
• /
• 2018

Stock price prediction has been a difficult problem to solve. There have been many studies to predict stock price scientifically, but it is still impossible to predict the exact price. Recently, a variety of types of cryptocurrency has been developed, beginning with Bitcoin, which is technically implemented as the concept of distributed ledger. Various approaches have been attempted to predict the price of cryptocurrency. Especially, it is various from attempts to stock prediction techniques in traditional stock market, to attempts to apply deep learning and reinforcement learning. Since the market for cryptocurrency has many new features that are not present in the existing traditional stock market, there is a growing demand for new analytical techniques suitable for the cryptocurrency market. In this study, we first collect and process seven cryptocurrency price data through Bithumb's API. Then, we use the gradient boosting model, which is a data-driven learning based machine learning model, and let the model learn the price data change of cryptocurrency. We also find the most optimal model parameters in the verification step, and finally evaluate the prediction performance of the cryptocurrency price trends.

• Food Science and Preservation
• /
• v.10 no.1
• /
• pp.28-31
• /
• 2003

For safe storage of agricultural products in the cold storage house, accurate monitoring of temperature, humidity and gas conditions is necessary. This study was conducted to develop an environmental monitoring and warning system for the cold storage house to improve safety of storage. The system developed in this study is able to monitor temperature, humidity and $CO_2$concentration in the storage house and to send alarm signal to the farmer by telephone and beeper when abnormal conditions have been occurred in the storage house. And the developed system use internet network so we can supervise storage conditions in the home. From the results of the performance test, it was found that the temperature and relative humidity can be controlled within the range of 0.5$^{\circ}C$ and $\pm$2 percent. And farmer's response was fair.

• Food Science and Preservation
• /
• v.17 no.2
• /
• pp.267-274
• /
• 2010

To obtain basic information on the potential use of Camellia japonica leaf as a raw material in functional food, leaf antioxidant and anticancer activities were investigated. The radical-scavenging activity of various solvent fractions from the leaf, as shown by the DPPH radical test, increased in a dose-dependent manner, with the water fraction showing the highest activity. The reducing power of various solvent fractions from the leaf was also dose-dependent, and, again, the water fraction showed the highest reducing power. The water fraction showed strong antioxidant activity in the linoleic acid test and was also capable of scavenging nitrite in a dose-dependent manner. Proportions of 92.15% and 95.61% of available nitrite were scavenged by the water and butanol fractions, respectively, at levels of $1,000{\mu}g/mL$. Both butanol and water fractions exhibited strong inhibitory effects on the growth of human lung and colon cancer cells. The total phenolic contents of the butanol and water fractions were 216.26 mg/g and 220.68 mg/g, respectively. High-performance liquid chromatography (HPLC) showed that quercetin and epicatechin were the predominant phenolic compounds in the water fraction. The activities of this fraction are attributable to the presence of these phenolic compounds, particularly quercetin and epicatechin.