• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.4
• /
• pp.143-152
• /
• 2021

It is very important to learn behavioral policies that allow multiple agents to work together organically for common goals in various real-world applications. In this multi-agent reinforcement learning (MARL) environment, most existing studies have adopted centralized training with decentralized execution (CTDE) methods as in effect standard frameworks. However, this multi-agent reinforcement learning method is difficult to effectively cope with in a dynamic environment in which new environmental changes that are not experienced during training time may constantly occur in real life situations. In order to effectively cope with this dynamic environment, this paper proposes a novel multi-agent reinforcement learning system, C-COMA. C-COMA is a continual learning model that assumes actual situations from the beginning and continuously learns the cooperative behavior policies of agents without dividing the training time and execution time of the agents separately. In this paper, we demonstrate the effectiveness and excellence of the proposed model C-COMA by implementing a dynamic mini-game based on Starcraft II, a representative real-time strategy game, and conducting various experiments using this environment.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.8
• /
• pp.1811-1819
• /
• 2013

In this paper, we carried out a study on how to find an appropriate helper node for cooperative communications, the role of which is very import to enhance system throughput of wireless communication system. The busy tone cooperative MAC (BT-COMAC) protocol proposed in this paper is a new cooperative MAC protocol with a reactive helper node scheme and maximizes the benefits of existing schemes while making up for their shortcomings. We conducted performance evaluation of this new protocol using computer simulation experiment. System throughput in bps and channel access delay are utilized as performance measures. We used a random way point mobility model where every communication node moves independently one another, and slow fading channel where every communication node decided its transmission rate with received power basis. Numerical results show that the new MAC protocol enhances system throughput as much as 15% of the existing scheme.

• Journal of KIISE
• /
• v.42 no.3
• /
• pp.399-407
• /
• 2015

Wireless sensor networks are used for data collection and processing from the surrounding environment for various applications. Since wireless sensor nodes operate on low computing power, restrictive battery capacity, and low network bandwidth, their architecture model has greatly affected the performance of applications. If applications have high computation complexity or require the real-time processing, the centralized architecture in wireless sensor networks have a delay in data processing. Otherwise, if applications only performed simple data collection for long period, the distributed architecture wasted battery energy in wireless sensors. In this paper, the energy consumption and processing delay were analyzed in centralized and distributed sensor networks. In addition, we proposed a new hybrid architecture for wireless sensor networks. According to the characteristic of applications, the proposed method had the optimal number of wireless sensors in wireless sensor networks.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.10 no.11
• /
• pp.3183-3192
• /
• 2009

The purpose of this paper is to introduce Naming application model for sensor networks. Currently, sensor networks comprised of sensor nodes have provided an application range which could not function before. However, unlike general network, current sensor networks are designed to cooperate with major wireless-capable sensor devices with limited resources. Thus, exporting/importing between individual sensor and current sensor networks is very inefficient and unstable. Attribute, schema and DIT(Directory Information Tree) must be designed for sensor network using SN LDAP application model in order to maintain transparency and provide constant service in a situation of data defect. With the system explained as above, Naming application model is made to manage SN Fuzzy Query. It shall be more efficient and stable structure as long as Naming application using a virtual equation in a certain environment with information collected from sensor node is provided. In this paper, I would like to introduce SN Fuzzy LDAP model for sensor network by quick Naming method. Also, naming application which is possible for fuzzy query in a certain environment based on the system will be proved.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.4
• /
• pp.222-227
• /
• 2022

In this paper, we implemented a Korean text generation and classification model based on a deep learning algorithm that can be applied to various industries. It consists of two implemented GAN-based Korean handwriting generation models and CNN-based Korean handwriting classification models. The GAN model consists of a generator model for generating fake Korean handwriting data and a discriminator model for discriminating fake handwritten data. In the case of the CNN model, the model was trained using the 'PHD08' dataset, and the learning result was 92.45. It was confirmed that Korean handwriting was classified with % accuracy. As a result of evaluating the performance of the classification model by integrating the Korean cursive data generated through the implemented GAN model and the training dataset of the existing CNN model, it was confirmed that the classification performance was 96.86%, which was superior to the existing classification performance.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.442-445
• /
• 2023

최근 딥러닝 및 로봇기술의 발전으로 인해 대량의 데이터를 빠르게 수집하고 처리하는 연구 분야들로 확대되었다. 이와 관련된 한 가지 분야로써 다중 로봇을 이용한 분산학습 연구가 있으며, 이는 단일 에이전트를 이용할 때보다 대량의 데이터를 빠르게 수집 및 처리하는데 용이하다. 본 연구에서는 기존 Distributed Neural Network Optimization (DiNNO) 알고리즘에서 제안한 정적 분산 학습방법과 달리 단계적 분산학습 방법을 새롭게 제안하였으며, 모델 성능을 향상시키기 위해 원시 변수를 근사하는 단계수를 상수로 고정하는 기존의 방식에서 통신회차가 늘어남에 따라 점진적으로 근사 횟수를 높이는 방법을 고안하여 새로운 알고리즘을 제안하였다. 기존 알고리즘과 제안된 알고리즘의 정성 및 정량적 성능 평가를 수행하기 MNIST 분류와 2 차원 평면도 지도화 실험을 수행하였으며, 그 결과 제안된 알고리즘이 기존 DiNNO 알고리즘보다 동일한 통신회차에서 높은 정확도를 보임과 함께 전역 최적점으로 빠르게 수렴하는 것을 입증하였다.

• The Journal of Information Systems
• /
• v.11 no.1
• /
• pp.75-103
• /
• 2002

현대사회는 전통적인 경제체제에서 디지털 경제체제로의 급속한 이전과 더불어 정보기술이 국가발전의 주요 원동력으로 등장하게 되었다. 특히, e-정부의 출현은 기존의 정부서비스에 대한 시공간적 제약을 극복하는 새로운 패러다임으로 간주되고 있다. 본 연구는 e-정부 실현을 위한 대표적인 사례로 KICIT을 중심으로 정보기술의 활용사례와 더불어 EC-Bank를 중심으로 한 지방정부의 사이버 마케팅에 대한 사례연구를 통해 e-정부의 조기 실현을 위한 산업부문의 e-비즈니스화에 대한 하나의 대안으로 소개하였다. e-구미의 모형은 행정, 생활, 산업 차원에서의 네트워크화, 균형과 조화, 부문간의 연계로 시행정의 디지털화, 사이버 정보생활의 실현 및 산업의 e-비즈니스화의 실현을 중심으로 제시하였다. 본 연구에서 KICIT은 중소도시에 집중되어 있는 지역정보센터의 산학관 협력의 모델케이스로 효율적으로 운영되고 있음을 확인할 수 있었다. 이제는 구미지역 기업체와 연구교육기관에서 핵심적인 지원기관으로 지역관련기관과의 보다 유기적인 협력을 계속 유지. 발전해 나가야 할 것이다. 지자체의 사이버 마케팅 사례인 EC-Bank는 정보서비스를 단순히 제공하는 수준인 전자상거래의 초기단계로 사이트 및 콘텐츠 관리의 소홀로 인하여 초기의 목적을 거의 이루지 못한 실정이다. 앞으로 산학관의 친밀한 협력과 e-프로세스의 대혁신을 통해서 새로운 업무활용, 웹기반의 거래처리 및 e-비즈니스 모형의 창출을 위해 계속적으로 노력해야할 것이다.

• Korean Journal of Construction Engineering and Management
• /
• v.11 no.3
• /
• pp.134-144
• /
• 2010

In China, the enhancement of water services has become the most crucial issue confronted with the rapid urbanization and industrialization process. A huge financial gap to meet the demand for water infrastructure and need for adopting advanced operation technology precipitated the rapid growth of PPP over the last 10 years. Diverse schemes of PPP such as TOT, Divestiture, and Management Contract and Lease have been practiced. Local governments and private investors/operator have adjusted their objectives and strategies to avoid potential pitfalls behind BOT projects in China. However, current academic research outcomes do not properly reflect important issues of BOT projects or related case studies in China. This limitation has brought in the lack of assessment of important risks and success factors required for the improvement of the body of risk management. In this regard, this study uses the market analysis method to identify major schemes of PPP water projects and conducts case studies on five PPP projects to identify key risk and success factors in association with each different scheme. It is expected that the risk and success factors identified from the cases will be used as reference to Korean companies which plan to enter the Chinese water market.

• Korean Journal of Construction Engineering and Management
• /
• v.19 no.5
• /
• pp.10-21
• /
• 2018

This study analyzes difficulties of enterprise activities in Chinese construction market by surveys on early entrants and interviews with experts. This study also suggests future strategy to enter the market by using inducement coefficient model. Korean construction companies in China are under heavy pressure to maintain requirement of licenses, despite recent deregulation. They are in dire predicament for market entry due to the Foreign Direct Investment System. It is almost impossible to participate in public projects and also it is not easy to do PPP projects. Therefore, Korean government should make more efforts to solve those issues through negotiations in FTA and GPA. For future expansion in Chinese construction market, it is highly recommended to boost cooperation models between Korea and China according to the empirical results of inducement coefficient model. Korean companies should collaborate with Chinese companies in some fields: smart city, environment and water treatment. Also, Korean government should support Korean companies by diplomatic means such as requesting for further opening of China's market. In GPA or GATS negotiation, Korean government should ask Chinese government that Korean companies can obtain order independently (without joint venture with Chinese companies) in China. Lastly, Korean construction companies should participate in construction projects ordered by international organizations such as ADB, AIIB.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.05c
• /
• pp.1805-1808
• /
• 2003

웹 서비스의 위치와 컨텐츠에 대한 정보를 제공하는 UDDI 는 현재 IBM, Microsoft 그리고 SAP 에서 공용 UDDI 레지스트리 서비스 형태로 제공되고 있다. 기존의 UDDI 레지스트리 서비스는 웹 서비스 사용자 입장에서 볼 때 이용이 어렵고, 제공업체에 따른 UDDI 의 검색 도구와 방식의 차이로 인한 혼란이 있으며, 검색 효율도 만족스럽지 못한 실정이다. 이에 본 논문에서는 이러한 문제를 보완하기 위해 UDDI 레지스트리 서비스 중 검색을 위한 에이전트 개념 모델을 제안한다 UDDI 검색 에이전트는 사용자와 공용 UDDI 레지스트리 서비스 사이에 위치하며, UDDI 서비스 이용의 편의를 위해 사용자의 정보를 관리하는 UDDI 브라우저, 필요 시 검색 과정의 각 상황에 대한 정보 및 해결책을 제공하는 Help Desk, 사용자의 UDDI 레지스트리 검색 성향을 분석하는 패턴 분석기, 그리고 이전에 검색한 웹 서비스 정보 검색 경로를 기록한 Search Map으로 구성된다. 이 외에 Search Map을 작성하기 위한 Search Map Building Engine 또한 사용된다 이러한 UDDI 검색 에이전트 개념 모델의 구성요소들은 서로 메시지를 주고받으며 유기적으로 협력해, 사용자 입장에서 보다 더 쉽고, 용이하고, 효율적인 UDDI 검색 서비스를 제공한다.