베이즈주의, 공약불가능성, 이론 변화

허원기 ( Won Ki Her ) 한국과학철학회 2015 과학철학 Vol.18 No.2

In this paper, I try to show that Bayesianism is a useful tool to analyze theory changes in scientific revolutions. For this purpose, first, I present a sketch of Salmon`s Bayesian reconstruction of theory change in scientific revolutions. Second, I examine and refute the Farmakis` skeptical argument based on the incommensurability thesis. Third, I point out the weaknesses of Yeo`s paradigm-dependent Bayesian Algorithm(PBA), and then I suggest another way to deal with the threat of incommensurability. Finally, I analyze the chemical revolution case by means of the Bayesian Algorithm(BA). This case study shows that the incommensurability thesis could not defeat Bayesianism.

피아제의 명제논리와 베이지언 인과추론에 근거한 구조구성주의 융합교수학습모형 개발

은은숙 새한철학회 2020 哲學論叢 Vol.99 No.1

With this article, I would like to introduce the new teaching-learning model synthesizing the Bayesian probabilistic models and Piaget’s propositional ones of learning. I call this new model “Bayesian structure-constructivist Model of Teaching-learning”(abbreviation: BMT). Teaching theory is usually based on analyzing the learning theory. I will first compare and criticise Piagetian learning theory and Bayesian one in terms of historical genesis and formalization. Then, I will present and justifier the main features of BMT. BMT first underlines that learning is an interaction between representation and sampling, that children always distinguish between week sampling and strong sampling, spacially between pedagogical sampling and non-pedagogical. Second, BMT supposes that, to select examples to facilitate learning, teachers have to use the Bayesian mode of pedagogical reasoning, and combine the direct instruction and discovery learning. Finally, Even though children may sometimes learn abstract rules earlier than more concrete ones(the blessing of abstraction), we should accept Piagetian logical grammars as a possible broad language of thought and hierarchical systems of logical levels. This is because the blessing of abstraction is consistent with developmental data: learning more abstract knowledge in most areas depends on first learning more concrete kinds of knowledge. 본 연구의 목적은 최근 20여 년 동안 진행되어 온 학습이론에 대한 피아제의 명제논리학적 학습이론과 베이즈주의의 확률론적 학습이론의 융합에 근거하는 새로운 융합교수학습모형을 개발하는 것이다. 연구자는 이 새로운 교수학습모델을 “베이지안 구조구성주의 교수학습모형”(Bayesian structure-constructivist Model of Teaching-learning: 이하 약칭 BMT)이라 명명한다. 본고는 역사-비판적 관점 및 형식화적 관점에서 피아제의 명제논리학적 학습모형에서 해석된 학습이론과 베이즈주의의 확률론적 추론모형에서 해석된 학습이론을 일차적으로 분석하고, 논문의 후반부에서는 이를 근거로 교수법의 관점에서 양자의 학습이론을 통합하는 새로운 교수학습모델, 즉 BMT의 중요한 특성들을 세부적으로 제시한다. 몇 가지 핵심만 언급하면, 첫째로, BMT는 개념 학습이든 이론 학습이든 모든 학습은 표상과 샘플링의 상호작용임을 강조한다. 매우 이른 나이에서부터 아동은 학습하기 이전에 자기에게 주어진 예시들이 약한 샘플링인지 강한 샘플링인지, 특히 교육학적 샘플링인지 비교육학적 샘플링인지를 구분한다. 둘째로, 따라서 학습을 촉진시킬 수 있는 샘플을 선택하기 위해서 교사는 교육학적 추론을 위한 베이지안 모델을 활용하면서 직접적인 강의식 수업과 발견학습 수업을 병행해야 한다. 셋째로, 아동은 때때로 구체적인 규칙들보다는 추상적인 규칙들을 먼저 학습하기도 하지만, 대부분의 영역에서 추상적인 지식의 학습은 구체적인 지식의 학습에 의존한다. 따라서 성공적인 학습을 위해서는 피아제가 강조한 보편적 언어로서 논리적 문법과 그들 사이의 위계성을 고려하는 샘플을 개발하고 제시해야 한다.

진화론에서의 확률에 대한 베이즈적 해석: 유전자 부동 모형의 경우

료타모리모토 ( Ryota Morimoto ) 한국과학철학회 2013 과학철학 Vol.16 No.2

The concept of probability is an integral part of evolutionary theory. What does probability represent If it represents the reality of the biological world, then it means that the world is indeterministic. In the classical world view, however, the probabilities appearing in the scientific context are interpreted as our ignorance of the deterministic world: they do not represent the real world. Alex Rosenberg (1994) argues that the probabilities used in evolutionary theory shouldn`t be interpreted realistically because they only reflect our ignorance of details. In this paper, I propose an alternative to Rosenberg`s interpretation. First, I give a critical appraisal of his arguments and show that the probabilities reflect not merely our ignorance but some aspects of reality. I discuss this issue by using genetic drift model as an example. Second, I compare evolutionary theory with statistical mechanics and show that in evolutionary theory we can update the probabilities rationally depending on what we know. Then I suggest that probability in evolutionary theory can be interpreted in Bayesian terms.

베이지안 의사결정 이론에 기반한 항공기 전기계통 신뢰성 분석

박경욱,이두열 대한기계학회 2025 대한기계학회논문집A Vol.49 No.7

항공기 운용 시 비행 안전은 매우 중요한 요소지만 경제성 또한 고려되어야 한다. 항공사는 표준 최소 장비 목록(MMEL)을 바탕으로 최소 장비 목록(MEL)을 작성하고 MEL을 기준으로 항공기 일부 장비가 부작동하는 상황에서도 비행을 지속한다. 그러나 MMEL은 시스템 부작동에 따른 위험 확률 및 영향 등은 제시하지 않기 때문에 운용자가 MMEL만으로 MEL을 작성하기에는 제한이 있다. 본 연구는 베이지안 네트워크를 바탕으로 항공기 전기계통 일부 부작동 상황에서 비행에 따른 위험을 정량적으로 분석하였다. 이후 베이지안 의사결정 이론을 바탕으로 불확실한 상황에서 최적의 의사결정 방안에 관해 연구하였다. 이를 통해 항공기 주요 계통 일부 부작동 상황에서 비행에 따른 위험을 정량적으로 평가할 수 있었고, 항공기 운용 중 불확실한 상황에서 의사결정 시 합리적인 판단 기준을 제시할 수 있었다. In aircraft operations, flight safety is one of the most crucial factors; however, it must be contemplated alongside cost-effectiveness. Therefore, airlines make their own minimum equipment list (MEL) based on the master minimum equipment list (MMEL), allowing for flights even when some equipment malfunctions. However, the MMEL does not provide information on the probability and impact of system malfunctions, which presents challenges in the process of preparing the MEL for systems that have a significant impact on flight safety. Thus, in this study, we analyzed the system malfunctions specified in the MMEL that affect flight safety using aircraft electrical system failure data. Bayesian networks were used to analyze the failure probability and risk of electrical systems, and the results were compared with aviation safety regulations to examine the impact of electrical system malfunctions on flight safety. Subsequently, we conducted a study on improving decision-making in uncertain situations based on Bayesian decision theory. The study enabled quantitative evaluation of the risk of flight operations in the event of a partial system malfunction in an aircraft, providing objective criteria for decision-making under uncertain conditions during aircraft operation.

베이지안 이론을 이용한 타입강관말뚝의 신뢰성 평가

박재현,김동욱,곽기석,정문경,김준영,정충기 한국지반공학회 2010 한국지반공학회논문집 Vol.26 No.7

For the development of load and resistance factor design, reliability analysis is required to calibrate resistance factors in the framework of reliability theory. The distribution of measured-to-predicted pile resistance ratio was obrained based on only the results of load tests conducted to failure for the assessment of uncertainty regarding pile resistance and used in the conventional reliability analysis. In other words, successful pile load test (piles resisted twice their design loads without failure) results were discarded, and therefore, were not reflected in the reliability analysis. In this paper, a new systematic method based on Bayesian theory is used to update reliability indices of driven steel pipe piles by adding more proof pile load test results, even not conducted to failure, to the prior distribution of pile resistance ratio. Fifty seven static pile load tests performed to failure in Korea were compiled for the construction of prior distribution of pile resistance ratio. The empirical method proposed by Meyerhof is used to calculate the predicted pile resistance. Reliability analyses were performed using the updated distribution of pile resistance ratio. The challenge of this study is that the distribution updates of pile resistance ratio are possible using the load test results even not conducted to failure, and that Bayesian updates are most effective when limited data are available for reliability analysis.

베이지안 네트워크를 활용한 교육 및 심리검사의 문항분석

강태훈(Tae hoon Kang) 한국교육평가학회 2014 교육평가연구 Vol.27 No.4

본 연구에서는 심리측정학 분야에서 활용 가능한 인지진단 도구의 하나로서 관심을 받고 있는 베이지안 네트워크를 통하여 교육 및 심리검사 실시 결과에 대한 양적 문항분석 작업을 수행하는 방법을 제시하고자 하였다. 이를 위하여 우선 베이지안 네트워크를 검사 자료에 적용할 때 추정되는 조건부 확률표의 수치들을 활용하여 문항곤란도와 문항변별도를 정의하였다. 그리고 측정하고자 하는 피험자 능력 측면에서 각각 일차원성과 다차원성을 가정하는 두 종류의 실제 검사 자료에, 베이지안 네트워크를 이용한 문항분석을 실시한 뒤, 고전검사이론과 문항반응이론을 적용하여 산출된 문항분석 결과와 비교하였다. 그 결과, 문항분석 차원에서 세 가지 접근 방법 간에 상당한 수준으로 일치하는 결과를 얻을 수 있었다. 이러한 결과를 통하여, 베이지안 네트워크를 활용한 문항분석이 주어진 검사 자료에 대한 문항분석을 실시함에 있어서 실제적으로 적용 가능한 방법이라고 말할 수 있었다. 추가로, 베이지안 네트워크 문항분석이 보다 풍부한 정보를 제공하면서 더 다양한 맥락에서 활용되기 위하여 앞으로 심층적으로 연구되어야 할 부분에 대하여 논의하였다. This study is to propose a quantitative item analysis method using the Bayesian network approach which is receiving attention as one of many possible cognitive diagnostic tools. To this end, the figures in conditional probability tables which are estimated through the Bayesian network are used to define item parameters such as item difficulty and item discrimination. Then, after performing the item analysis jobs for two kinds of actual data sets with the Bayesian network, the results were compared to those produced by the applications of classical test theory and item response theory. As a result, in terms of quantitative item analysis, similar results among the three approaches could be obtained. Through these results, it can be seen that the Bayesian network could be another applicable method for the purpose of item analysis. Finally, the future study issues to make this item analysis method provide richer information and be utilized in more complicated contexts were further discussed.

Investigation of the Probability of a Safe Evacuation to Succeed in Subway Fire Emergencies Based on Bayesian Theory

Jinghong Wang,Wenyu Yan,Han Xu,Youran Zhi,Zhirong Wang,Juncheng Jiang 대한토목학회 2018 KSCE Journal of Civil Engineering Vol.22 No.3

Emergency evacuation in subway fires is one of the hot issues in public safety. Based on Bayesian theory, a dynamic evacuation risk analysis model considering both psychological and behavioral responses of evacuees was proposed in this contribution. A subway fire scene was modeled using the Event Tree Analysis (ETA) according to the questionnaire survey at Nanjing Xinjiekou station. In this fire scenario, three control events (the fire alarm, the exhaust system and the evacuation route) were related to fire evacuation control and two control events (active escape and passive escape), which were specifically defined, were related to personnel characteristics. Subsequently, the probability of the final state of each scene was calculated by Monte Carlo simulation. Combined with the statistics of accidents and safe evacuation of subway fire cases from 1976 to 2013, the posterior probability distribution of safety evacuation was obtained after real-time dynamic updating of its precursor probability through Bayesian theory. Results show that the posterior probability of safety evacuation to succeed is around 0.8. This indicates that there is still a probability of 0.2 resulting in injury or fatalities in subway accidents even though most passengers are safely evacuated. It also suggests that utilization of 0.2 as the safety threshold would be an appropriate choice.

피아제의 명제논리적 학습이론과 베이즈주의 인과추론적 학습이론 사이의 논쟁 넘어서기

은은숙(Eun, Eun-Suk) 새한철학회 2019 哲學論叢 Vol.97 No.3

15년 전부터 등장한 베이지안 확률론적 추론 모형은 통계학, 과학철학, 심리학, 인지과학, 컴퓨터과학, 신경과학 등에서 학계의 연구를 지배하는 강력한 핵심논제가 된 후로 이젠 교육학과 심지어는 논리학 분야에까지 큰 반향을 일으키고 있다. 이러한 다양화로 인해, 굿에 따르면, 베이즈주의 유형은 46,656 가지로 분류된다. 이러한 다양한 모형들 중에서 베이지안 확률모형을 학습이론에 적용하려는 학자들은 학습자의 학습 과정이 정확히 베이지안 확률추론 과정을 항상 따른다고 주장한다. 그런데 이러한 확률론적 모형은 피아제의 구성주의적 전망을 계승한 것은 분명하다. 왜냐하면, 이 모델을 지지하는 학자들 자신이 자신들의 학적 운동을 “새로운 합리적 구성주의”라고 명시적으로 명명하기 때문이다. 그럼에도 불구하고, 이러한 확률론적 학습모형은 “지식은 인과성을 중심으로 구조화된다.”라는 가설을 확고하게 지지하면서, 피아제의 명제 논리적 모형에 의한 학습이론을 강하게 비판한다. 그러나 베이지안 학습 이론가들은 경험을 통한 새로운 항들의 연결로서의 학습, 그리고 학습에서 모순 및 인과성의 역할 등에 대한 피아제의 세부적인 연구 성과들을 놓치고 있다. 다른 한편으로 베이지안 학습이론은 여러 가지 점에서 온전한 발달 이론을 제공하기에는 아직 미흡하다. 이런 점에서, 필자는 현상 없이는 구조를 이해할 수 없듯이, 구조 없이는 현상을 이해할 수 없다는 사실을 강조하면서, 인과성 추론을 핵심으로 하는 베이지안 학습 모델과 인과적 물리적 인식뿐만 아니라, 동시에 논리-수학적 인식 모델을 강조하는 피아제의 학습 모델은 새로이 종합되어야 한다는 것을 논증하고자 한다. For more than 15 years, Bayesian probabilistic models have been dominant ones not only in the field of statistics, computational sciences, and machine learning, but also in epistemology, philosophy of sciences, and in cognitive sciences. This revolution extended in pedagogical circle and logics has tried to offer a renewal of the theories of learning, especially criticizing Piagetian theory. According to Bayesian learning theory, almost all our knowledge is only probable, and the core way of arriving at the truth, such as induction and analogy, is based on probabilities. But, in my opinion, the supporters of probabilistic models was so quick that they could not have the enough time to explore the rich works of Piaget. So, I will reexamine and compare the works of Piaget with those of the probabilistic models. I will show that the back and forth movement between data and hypotheses of Bayesian models is coherent with the learning through experience and especially contradiction which are important in Piaget" dynamic theory of assimilation and accommodation. I will then suppose that the probabilistic models are not enough to provide the global developmental models, such as the dynamic systems perspective attempting to offer the circular causality that governs part-part, part-whole, and whole-part relations. I will finally emphasize that the probabilistic models and Piagetian ones of logico-mathematical operations should be synthesized.

베이즈주의와 과학혁명: 새로운 이론과 비정밀 확률주의

박일호 ( Ilho Park ) 한국논리학회 2022 論理硏究 Vol.25 No.2

According to John Earman, Bayesianism can regard Kuhnian scientific revolutions as introductions of new theories to the existing belief set. I will assume in this paper that this view basically on the right track. However, it will be argued that some other works like Famakis (2008), Yeo (2011), Her (2015), and Chun (2011, 2019), which hinge on, or are related to Earman’s view, have some problems. Moreover, I will suggest a new Bayesian way of modeling Kuhnian scientific revolutions and so the introduction of new theories. In particular, it will be argued that imprecise probabilism, which represents an agent’s belief state by means of a set of probability functions, is a good alternative to a more standard approach.

A Particle Swarm Optimization and Immune Theory-Based Algorithm for Structure Learning of Bayesian Networks

Xiao-Lin Li 보안공학연구지원센터 2010 International Journal of Database Theory and Appli Vol.3 No.2