• The Journal of Korean Institute for Practical Engineering Education
• /
• v.5 no.1
• /
• pp.1-13
• /
• 2013

For a basic engineering education the confirmation and verification of the deductive Inference was studied and the principle of probability inference was applied. The background of introduction of deductive Inference and its test method was mentioned, and historic arguments on the compatibility of deductive statistical inference was summarized and analyzed. Philosophical arguments on the deductive confirmation for engineering experiments was introduced. Premise, procedure, and control of the experiments are studied. As an example of the deductive probability inference three groups of experimental data were used in order to find successful inferences respectively.

• Journal of Intelligence and Information Systems
• /
• v.1 no.1
• /
• pp.1-22
• /
• 1995

본 연구에서 퍼지인식도(Fuzzy Cognitive Map) 개념을 기초로 하여 (1) 특정 문제영역에 대한 전문가의 인과관계 지식(causal knowledge)을 추출하는 알고리즘을 제시하고, (2) 이 알고리즘에 기초하여 작성된 해당 문제영역에 대한 여러 전문가들의 인과관계 지식을 계층별로 분해하여, (3) 해당 계층간의 양방향 추론이 가능한 추론메카니즘을 제시하고자 한다. 특정 문제영역에 있어서의 인과관계 지식이란 해당 문제를 구성하는 여러 개념간에 존재하는 인과관계를 표현한 지식을 의미한다. 이러한 인과관계 지식은 기존의 IF-THEN 형태의 규칙과는 달리 행렬형태로 표현되기 때문에 수학적인 연산이 가능하다. 특정 문제영역에 대한 전문가의 인과관계 지식을 추출하는 알고리즘은 집합연산에 의거하여 개발되었으며, 특히 상반된 의견을 보이는 전문가들의 의견을 통합하여 하나의 통합된 인과관계 지식베이스를 구축하는데 유용하다. 그러나, 주어진 문제가 복잡하여 다양한 개념들이 수반되면, 자연히 인과관계 지식베이스의 규모도 커지게 되므로 이를 다루는데 비효율성이 개재되기 마련이다. 따라서 이러한 비효율성을 해소하기 위하여 주어진 문제를 여러계측(Hierarchy)으로 분해하여, 해당 계층별로 인과관계 지식베이스를 구축하고 각 계층별 인과관계 지식베이스를 연결하여 추론하는 메카니즘을 개발하면 효과적인 추론이 가능하다. 이러한 계층별 분해는 행렬의 분해와 같은 개념으로도 이해될 수 있다는 특징이 있어 그 연산이 간단명료하다는 장점이 있다. 이와같이 분해된 인과관계 지식베이스는 계층간의 추론메카니즘을 통하여 서로 연결된다. 이를 위하여 본 연구에서는 상향 또는 하향방식이 추론이 가능한 양방향 추론방식을 제시하여 주식시장에서의 투자분석 문제에 적용하여 그 효율성을 검증하였다.

• Annual Conference of KIPS
• /
• 2012.11a
• /
• pp.833-836
• /
• 2012

외부에서 특정 네트워크의 방화벽 정책을 추론하기 위해서는 Active Probing 을 이용한 탐지 패킷의 응답을 분석하여야 한다. 하지만, 외부에서 특정 네트워크로 탐지 패킷을 어떻게 전송하는가에 따라 방화벽에서 네트워크 공격으로 탐지되기 때문에 무분별하게 탐지 패킷을 전송하는 방법은 위험하다. 본 논문에서는 방화벽 장비가 Active Probing 을 이용한 방화벽 정책 추론 방법에 어떠한 영향을 주는지에 대해서 분석한다. 그리고 실제 방화벽 정책과 추론된 방화벽 정책을 비교하여 방화벽 정책 추론 방법의 정확성을 검증한다.

• (The) Korean Journal of Educational Psychology
• /
• v.22 no.2
• /
• pp.367-384
• /
• 2008

The purpose of this study was to develop picture testing instrument for measuring children's causal reasoning ability on events that can occur in daily life. The measurement instrument contains three domain of human development; haman behavior domain, human psychology domain, and natural/physical domain. Through this study, researchers designed a concept model based on theoretical framework and prior studies and investigated the reliability and validity of the measurement instrument which was developed in accordance with the concept model. For the empirical validation research, a pretest was conducted to 336 elementary school students in grade 2 to 4 in Seoul. Considering reliability and validity of the developed measurement instrument and factor loadings, researchers sorted out 18 questions. And then 18 question test and KICE Critical Thinking Ability Test was conducted to 509 elementary school students in grade 1 to 4 in Seoul. According to the result of the tests, the researchers sorted out final 12 questions. The Cronbach's alpha, reliability of the children's causal reasoning ability test consisted of the final 12 question, was significant as .72. Also, the result of exploratory factor analysis showed that factors of this test were haman behavior domain, human psychology domain, and natural/physical domain. Moreover, the correlation between the KEDI Reasoning Ability Test(2003) scores and the scores of the test developed in the current study was significant as .55. Finally, the result of the analysis about children's grade differences, the development by discrepancy of age was significant in total points and that of each domain. The children's causal reasoning ability test developed by this study can be useful not only for the evaluation of children's thinking ability but also for screening of the handicapped children in thinking ability development.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.4
• /
• pp.304-310
• /
• 2017

This study is a methodological research study that tests the validity and reliability of the NCRC (Nurse Clinical Reasoning Competence scale), an instrument developed by Liou and his colleagues as the basic data for enhancing the clinical reasoning competence of nurses, by translating it into Korean and checking the similarity of the sentence structure and meaning (between the two versions?). This study verified its validity and reliability by examining 166 nurses working in four tertiary hospitals located in Seoul and Busan. An analysis of the content validity by experts showed that all of the items have a content validity higher than CVI 0.8. From the exploratory and confirmatory factor analysis, it was found that the instrument includes a total of 15 items consisting of one factor. In addition, the correlation with the Korean version of the Nurse Clinical Reasoning Competence scale is confirmed to test the concurrent validity, by using a measurement tool of nurses' critical thinking dispositions and clinical decision-making abilities (correlation coefficient =.55-.64(p<.001) and Cronbach's ${\alpha}=.93$). Thus, the Korean version of the NCRC may be a useful instrument for evaluating the clinical reasoning competence of Korean nurses and providing the basic data for assessing their clinical reasoning competence and developing their promotion strategies.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.4
• /
• pp.239-250
• /
• 2013

Context-aware mobile computing has become the primary approach to realize automatic, autonomous, and user-centric computing in the context of largely increasing the amount of mobile devices used that embed available sensors. However, designing an inference engine nonetheless requires the tasks of analyzing contexts, situations that can be inferred, etc. Moreover, a mobile device has limited resources and limited computation capability, which results in recognizing the common sense of its unsuitable environment for processing inference. Hence, we propose context-situation reasoning elements and their formal models in this paper, and we verify the formal models' applicability by applying them to an example. Finally, we design and implement an inference engine that realize the context-situation inference elements in computing environment, and we experiment an example by using the proposed inference engine to verify applicability and reusability of the inference engine.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.12
• /
• pp.485-496
• /
• 2018

So far, the deep learning, a field of artificial intelligence, has achieved remarkable results in solving problems from unstructured data. However, it is difficult to comprehensively judge situations like humans, and did not reach the level of intelligence that deduced their relations and predicted the next situation. Recently, deep neural networks show that artificial intelligence can possess powerful relational reasoning that is core intellectual ability of human being. In this paper, to analyze and observe the performance of Relation Networks (RN) among the neural networks for relational reasoning, two types of RN-based deep neural network models were constructed and compared with the baseline model. One is a visual question answering RN model using Sort-of-CLEVR and the other is a text-based question answering RN model using bAbI task. In order to maximize the performance of the RN-based model, various performance improvement experiments such as hyper parameters tuning have been proposed and performed. The effectiveness of the proposed performance improvement methods has been verified by applying to the visual QA RN model and the text-based QA RN model, and the new domain model using the dialogue-based LL dataset. As a result of the various experiments, it is found that the initial learning rate is a key factor in determining the performance of the model in both types of RN models. We have observed that the optimal initial learning rate setting found by the proposed random search method can improve the performance of the model up to 99.8%.

• Journal of KIISE
• /
• v.42 no.12
• /
• pp.1590-1599
• /
• 2015

As a variety of personal medical devices appear, it is possible to acquire a large number of diverse medical contexts from the devices. There have been efforts to analyze the medical contexts via software applications. In this paper, we propose a generic model of medical analytics schemes that are used by medical experts, identify inference methods for realizing each medical analytics scheme, and present guidelines for applying the inference methods to the medical analytics schemes. Additionally, we develop a PoC inference system and analyze real medical contexts to diagnose relevant diseases so that we can validate the feasibility and effectiveness of the proposed medical analytics schemes and guidelines of applying inference methods.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.203-208
• /
• 2023

본 논문에서는 대규모 언어 모델의 인컨텍스트 러닝과 프롬프팅을 활용하여 수치 추론 태스크 데이터셋을 효과적으로 증강시킬 수 있는 방법론을 제안한다. 또한 모델로 하여금 수치 추론 데이터의 이해를 도울 수 있는 전처리와 요구사항을 만족하지 못하는 결과물을 필터링 하는 검증 단계를 추가하여 생성되는 데이터의 퀄리티를 보장하고자 하였다. 이렇게 얻어진 증강 절차를 거쳐 증강을 진행한 뒤 추론용 모델 학습을 통해 다른 증강 방법론보다 우리의 방법론으로 증강된 데이터셋으로 학습된 모델이 더 높은 성능을 낼 수 있음을 보였다. 실험 결과 우리의 증강 데이터로 학습된 모델은 원본 데이터로 학습된 모델보다 모든 지표에서 2%p 이상의 성능 향상을 보였으며 다양한 케이스를 통해 우리의 모델이 수치 추론 학습 데이터의 다양성을 크게 향상시킬 수 있음을 확인하였다.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2005.11a
• /
• pp.195-198
• /
• 2005

최근 실내 환경에서 영상 정보를 사용하여 로봇이 서비스를 제공하기 위한 연구가 활발하다. 과거 영상 처리 접근 방법은 산업 환경과 같은 예측 가능한 환경을 바탕으로 미리 정의된 기하학적 모델을 통해 상황을 인식하였기에, 이를 실내 환경과 같은 가변적인 환경에 적용할 시 성능이 저하된다. 이에 지식을 기반으로 불확실성을 해결하여 정확도를 향상 시킴으로써 영상 인식 성능을 높이기 위한 다양한 연구가 진행되어 왔다. 본 논문에서는 실내에서 활동하는 서비스 로봇의 물체인식 성능을 향상시키기 위해, 대상 물체가 다른 물체에 의해서 가려져 있는 경우 대상 물체의 존재 여부를 추론하기 위한 베이지안 네트워크 모델링 방법을 제안한다. 제안하는 방법은 작은 단위로 설계된 베이지안 네트워크들을 상황에 따라 결합하여 추론 모델이 구성되게 하였고 물체간의 관계를 효과적으로 표현하고 초기 확률 값을 단일하게 유지하기 위해 제안된 확률 값 설정 방법을 사용하였다. 실험은 물체 관계를 추론하는 모듈의 성능을 검증하기 위해 수행되었는데, 5가지 장소에서 82.8$\%$의 정확도를 보여주었다.