• 한국지능정보시스템학회:학술대회논문집
• /
• 한국지능정보시스템학회 2003년도 춘계학술대회
• /
• pp.319-328
• /
• 2003

With the proliferation of WWW, providing more intelligence to Web sites has become a major concern in e-business industry. In recent days, this trend is more accelerated by prosperity of CRM (Customer Relationship Management) in terms of various aspects such as product recommendation, self after service, etc. To accomplish this goal, many e-companies are eager to embed web enabled rule-based system, that is, expert systems into their Web sites and several well-known commercial tools are already available in the market. Most of those tools are developed based on CGI so far but CGI based systems inherently suffer over-burden problem when there are too many service demands at the same time due to the nature of CGI. To overcome this limitation of the existing CGI based expert systems, we propose a new form of Web-enabled expert system using hyperlink-based inference mechanism. In terms of burden to Web server, our approach is proven to outperform CGI based approach theoretically and also empirically. For practical purpose, our this approach is implemented in a software system, WeBIS and a graphic rule editing methodology, Expert Diagram is incorporated into the system to facilitates rule generation and maintenance. WeBIS is now successfully operated for financial consulting in the web site of a leading financial consulting company in Korea.

• Journal of the Korean Statistical Society
• /
• 제34권2호
• /
• pp.85-98
• /
• 2005

A Bayesian analysis for the product of different powers of k independent Poisson rates, written ${\theta}$, is developed. This is done by considering a prior for ${\theta}$ that satisfies the differential equation due to Tibshirani and induces a proper posterior distribution. The Gibbs sampling procedure utilizing the rejection method is suggested for the posterior inference of ${\theta}$. The procedure is straightforward to specify distributionally and to implement computationally, with output readily adapted for required inference summaries. A salient feature of the procedure is that it provides a unified method for inferencing ${\theta}$ with any type of powers, and hence it solves all the existing problems (in inferencing ${\theta}$) simultaneously in a completely satisfactory way, at least within the Bayesian framework. In two examples, practical applications of the procedure is described.

• Journal of the Korean Statistical Society
• /
• 제31권4호
• /
• pp.415-431
• /
• 2002

A unified framework for statistical inference for the mean of the normal distribution to derive point estimates, confidence intervals and statistical tests is proposed. This optimal design is justified after investigating the basic information and requirements that are possible and impossible to control when specifying practical and statistical requirements. Point estimation is only credible when viewed in the larger context of interval estimation, since the information required for optimal point estimation is unspecifiable. Triple sampling is proposed and justified as a reasonable sampling vehicle to achieve the specifiable requirements within the unified framework.

• 한국항해항만학회:학술대회논문집
• /
• 한국항해항만학회 2006년도 International Symposium on GPS/GNSS Vol.1
• /
• pp.301-306
• /
• 2006

The important field of research on ship operation is related to the high efficiency of transportation, the convenience of maneuvering ships and the safety of navigation. As a way of practical application for a smart ship based on network system, this paper proposes the intelligent support system for ship steering control system based on TCP/IP and desires to testify the validity of the proposal by applying the fuzzy control model to the steering control system. As the specific study methods, the fuzzy inference was adopted to build the maneuvering models of steersman, and then the network system was implemented using the TCP/IP socket-based programming. Lastly, the miniature model steering control system combined with LIBL (Linguistic Instruction-based Learning) was designed to testify for its effectiveness.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1993년도 하계학술대회 논문집 A
• /
• pp.214-217
• /
• 1993

This paper presents a new identification method which utilizes fuzzy inference in parameter identification. The proposed system has an additional control loop where a real plant is replaced by a plant model. The control system to be designed is to satisfy the following specifications: 1) It has zero steady-state error. 2) It has adequate damping characteristics. 3) 1),2) satisfied, it has a shortest rise-time. Fuzzy rules describe the relationship between comparison results of the features and magnitude of modification in the model parameter values. This method is effective in auto-tuning because the response of the closed loop is verified. The proposed method is tested in simulation for several plants with first- order lags and dead-times. The results show that the proposed method is effective in practical use.

• Journal of Information Processing Systems
• /
• 제18권1호
• /
• pp.12-25
• /
• 2022

In recent years, deep convolutional neural networks have made significant progress in the research of single image super-resolution. However, it is difficult to be applied in practical computing terminals or embedded devices due to a large number of parameters and computational effort. To balance these problems, we propose CSRNet, a lightweight neural network based on channel split residual learning structure, to reconstruct highresolution images from low-resolution images. Lightweight refers to designing a neural network with fewer parameters and a simplified structure for lower memory consumption and faster inference speed. At the same time, it is ensured that the performance of recovering high-resolution images is not degraded. In CSRNet, we reduce the parameters and computation by channel split residual learning. Simultaneously, we propose a double-upsampling network structure to improve the performance of the lightweight super-resolution network and make it easy to train. Finally, we propose a new evaluation metric for the lightweight approaches named 100_FPS. Experiments show that our proposed CSRNet not only speeds up the inference of the neural network and reduces memory consumption, but also performs well on single image super-resolution.

• 센서학회지
• /
• 제30권4호
• /
• pp.261-266
• /
• 2021

A simple memristor model is proposed for the neuromorphic system in the Simulation Program for Integrated Circuits Emphasis (SPICE). The memristive I-V characteristics with different voltage and frequencies were analyzed. And with the model, we configured a learning and inference system with 4 by 4 memristor array to show the practical use of the model. We examined the applicability by configuring the simplest neuromorphic circuit. The total simulation time for the proposed model was 18% lesser than that for the one-memristor model. When compared with more memristor models in a circuit, the time became even shorter.

• 실천공학교육논문지
• /
• 제6권1호
• /
• pp.1-7
• /
• 2014

이 연구는 귀납법적 확증(confirmation)을 위해서 실행하는 실험이 확률론적 추론임을 밝히고 이의 단점을 보완하기 위해서 적용된 확률적 확인론에 따라서 수행하는 실험에서 통제하는 절차와 방법을 다루었다. 이 통제란 알려진 조건과 상황들을 인위적으로 조작하되 가설에서 명시된 특정조건(즉 독립변수) 이외의 조건(기타 변수)이 작용하여 현상이 발생(종속변수)될 가능성을 막는 것이다. 특정조건에 의하여 현상이 발생할 가능성을 극대화 시키고 기타 조건에 의하여 현상이 발생할 가능성을 최소화시키는 것이다. 그렇게 함으로써 특정 선행사건(독립변수)과 후행사건(종속변수) 사이의 진정한 인과관계성(the causal relationship)을 객관적으로 정확하게 추론하도록 보장하는 방법이 통제인 것이다. 이 연구에서는 변수통제에 대해 좀 더 구체적으로 다루기 위해 대표적인 헬스기구인 Elliptical Trainer (ET)와 Tread Mill (TM)를 이용한 실험에서 변수통제를 가하여 운동속도(독립변수)에 따른 에너지 소비량 변화(종속변수)와 운동속도(독립변수)에 따른 호흡교환율(RER)과 심장 박동율(HR)의 변화(종속변수)를 얻었다. 마지막으로 이 통제를 통해 얻어진 실험결과들을 그래프로 그려 그 결과에 따라 연구의 결론이 다루어졌다.

• 전자공학회논문지CI
• /
• 제38권6호
• /
• pp.57-69
• /
• 2001

전문가 시스템은 인공지능의 한 분야로서 인간의 사고방식을 모방함으로써 다양한 분야에서 야기되는 문제들을 해결해준다. 대부분의 전문가 시스템은 추론엔진과 지식베이스등과 같은 많은 요소들로 구성된다. 특히 전문가 시스템의 성능은 추론엔진의 효율성에 의해 좌우된다 이러한 추론 엔진은 지식베이스가 구축될 때, 가능한한 적은 제약성을 가져야 함은 물론, 다양한 추론 방법을 제공해야 한다는 특정을 갖고 있어야 한다. 이 논문에서는 도메인 지식 표현을 위한 지식스키마와, 추론을 위한 지식구현 기법, 그리고 블랙보드와 추론엔진을 혼합한 지식베이스 엔진을 제안한다. 그리고 제안한 기법들을 이용하여 구축한 중공업 장비 진단 전문가 시스템에 대해서 설명한다. 이 논문에서 연구한 고장진단 지식기반 시스템은 지식기반 시스템 연구분야의 실질적 기반이 될 수 있을 것이다.

• 조명전기설비학회논문지
• /
• 제17권4호
• /
• pp.57-63
• /
• 2003

가공 송전선로의 아연도금 강심 알루미늄연선 도체들은 장기간 동안 대기오염에 의해 서서히 열화되었기 때문에, 2천년 대에 이르러 수많은 도체들이 예상된 유효수명을 초과하였다. 대부분의 도체들은 경제적인 운용 측면에서 현재 상태들을 평가하지 않으면 안되므로, 이 논문에서는 경년, 환경지표, 및 도체구조와 같은 중요 파라미터들을 사용하여 노화도체의 현재 상태를 평가하기 위한 방법을 제안하였다. 노화도체의 수명에 대응하는 열화등급을 예측하기 위한 진단 방법을 기술하였으며, 이 시스템은 전문가 지식과 경험을 토대로 적응 뉴로퍼지 추론시스템 (Adaptive Neuro-Fuzzy Inference System)으로 설계하였다. 이 진단시스템을 국내의 송전선로에 적용하여, 이 시스템이 노화 ACSR 도체를 비파괴적으로 진단하고 경제적으로 운용하기 위한 방안으로서 효과적으로 사용될 수 있음을 밝혔다.