• Proceedings of the Korean Information Science Society Conference
• /
• 2005.07b
• /
• pp.724-726
• /
• 2005

베이지안망(Bayesian network)은 다수의 변수들 사이의 확률적 관계(조건부독립성: conditional independence)를 그래프 구조로 표현하는 모델이다. 이러한 베이지안망은 비감독학습(unsupervised teaming)을 통한 데이터마이닝에 적합하다. 이를 위해 데이터로부터 베이지안망의 구조와 파라미터를 학습하게 된다. 주어진 데이터의 likelihood를 최대로 하는 베이지안망 구조를 찾는 문제는 NP-hard임이 알려져 있으므로, greedy search를 통한 근사해(approximate solution)를 구하는 방법이 주로 이용된다. 하지만 이러한 근사적 학습방법들도 데이터를 구성하는 변수들이 수천 - 수만에 이르는 경우, 방대한 계산량으로 인해 그 적용이 실질적으로 불가능하게 된다. 본 논문에서는 그러한 대규모 데이터에서 학습될 수 있는 계층적 베이지안망(hierarchical Bayesian network) 모델 및 그 학습방법을 제안하고, 그 가능성을 실험을 통해 보인다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2007.11a
• /
• pp.9-12
• /
• 2007

테이블 탑 디스플레이는 사람에게 친숙한 상호작용의 매개체인 손을 입력장치로 이용하는 일종의 탁자형 멀티 터치스크린이라고 할 수 있다. 본 논문에서는 이러한 환경에서 손가락 제스쳐를 활용하여 필기 숫자를 인식하는 연구를 수행함으로써 테이블 탑 디스플레이에 적합한 필기 숫자 인식 기술을 개발하였고, 이로 인해 추후 진행될 연속 숫자 혹은 특수기호의 성공적인 인식 가능성을 확인하였다. 실험 과정은 테이블 탑 디스플레이의 표면을 통해 입력된 손가락 궤적을 잡음제거, 대표점 추출등의 전처리 과정을 거쳐 16-방향 체인코드로 변환하고, 변환된 체인코드의 학습 및 필기 숫자 인식에 확률 통계적 모델인 은닉 마르코프 모델을 이용하였다. 학습에는 총 300개 필기 숫자 데이터를 이용하였고, 인식 실험에 사용한 별도의 100개의 필기 숫자 데이터에 대해 97%의 정인식율을 보였다.

• Annual Conference on Human and Language Technology
• /
• 2004.10d
• /
• pp.73-81
• /
• 2004

한국어에서의 품사 결정 문제는 형태론적 중의성 문제도 있지만, 영어에는 발생하지 않는 동품사 중의성 문제로 더 까다롭다. 이러한 문제들은 어휘 문맥을 고려하지 않고서는 해결하기 어렵다. 통계 자료 부족 문제에 쉽게 대처하는 모델이 필요하며 문맥에 따른 품사를 결정하고자 할 때 서로 다른 형태의 여러 가지 어휘 문맥 정보를 반영할 수 있는 모델이 필요하다. 본 논문에서는 이런 점에 가장 적합한 최대 엔트로피(maximum entropy : ME) 모델을 품사태깅 작업에 이용하는 문제에 대해 다룬다. 어휘 문맥 정보를 이용하기 위한 자질함수가 매우 많아지는 문제에 대처하기 위해 필요에 따라 어휘 문맥 정보를 사전화 한다. 본 시스템의 특징으로는 어절 단위 품사 태깅을 위한 처리 기법. 어절의 형태소 분석열에 대한 어절 내부 확률 계산. ME 모델의 정규화 과정 생략에 의한 성능 향상, 디코딩 경로의 확장과 같은 점들이 있다. 실험을 통하여 본 연구의 기법이 높은 성능의 시스템을 달성할 수 있음을 알게 되었다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2011.04a
• /
• pp.152-155
• /
• 2011

최근 사회생활의 다변화로 인한 개인의 질환을 예방하고 건강을 증진시키기 위한 개인 웰니스 관리는 현대 사회의 성인에게는 필수적인 자기 관리에 해당된다. 본 논문는 이러한 웰니스 관리에 적절한 상황 모델로써 상황 데이터를 추론할 수 있는 SWRL 상황규칙과 불확실성을 표현한 베이지안 네트워크를 포함한 통합 온톨로지 기반 상황모델을 제시하였다. 제안한 상황모델에 포함된 추론 규칙은 웰니스 관리에 필요한 상황 서비스를 수행하는 액션들을 정의한다. 즉 상황 온톨로지에 SWRL 규칙을 포함함으로써 주로 웹 시멘틱에 사용되고 있는 OWL 언어를 상황인식 분야의 지식 베이스 구축에도 적합하도록 하였다. 그리고 웰니스 관리를 위해 상황 온톨로지로 표현되는 원시 상황 데이터는 센서 부정확성, 또는 개인 판단기준 차이로 인해 불확실성을 포함하므로, 어떤 논리적 상황 데이터는 불확실성을 고려하여 추론되어야 하기 때문에 본 논문은 상황 온톨로지 및 SWRL 규칙과 함께 베이지안 네트워크를 함께 표현할 수 있게 하여 OWL 상황 온톨로지 기반 규칙 추론뿐만 아니라 확률 추론을 용이하게 하였다.

• The Journal of the Acoustical Society of Korea
• /
• v.27 no.8
• /
• pp.435-442
• /
• 2008

The n-gram model is appropriate for languages, such as English, in which the word-order is grammatically rigid. However, it is not suitable for Korean in which the word-order is relatively free. Previous work proposed a twoply HMM that reflected the characteristics of Korean but failed to reflect word-order structures among words. In this paper, we define a new segment unit which combines two words in order to reflect the characteristic of word-order among adjacent words that appear in verbal morphemes. Moreover, we propose a two-path language model that estimates probabilities depending on the context based on the proposed segment unit. Experimental results show that the proposed two-path language model yields 25.68% perplexity improvement compared to the previous Korean language models and reduces 94.03% perplexity for the prediction of verbal morphemes where words are combined.

• Journal of the Korean Geophysical Society
• /
• v.3 no.3
• /
• pp.161-174
• /
• 2000

Bayesian inversion is a stable approach to infer the subsurface structure with the limited data from geophysical explorations. In geophysical inverse process, due to the finite and discrete characteristics of field data and modeling process, some uncertainties are inherent and therefore probabilistic approach to the geophysical inversion is required. Bayesian framework provides theoretical base for the confidency and uncertainty analysis for the inference. However, most of the Bayesian inversion require the integration process of high dimension, so massive calculations like a Monte Carlo integration is demanded to solve it. This method, though, seemed suitable to apply to the geophysical problems which have the characteristics of highly non-linearity, we are faced to meet the promptness and convenience in field process. In this study, by the Gaussian approximation for the observed data and a priori information, fast Bayesian inversion scheme is developed and applied to the model problem with electric well logging and dipole-dipole resistivity data. Each covariance matrices are induced by geostatistical method and optimization technique resulted in maximum a posteriori information. Especially a priori information is evaluated by the cross-validation technique. And the uncertainty analysis was performed to interpret the resistivity structure by simulation of a posteriori covariance matrix.

• The KIPS Transactions:PartB
• /
• v.11B no.2
• /
• pp.233-240
• /
• 2004

This paper presents a method to implicitly resolve ambiguities using dynamic incremental clustering in Korean-to-English and Japanese-to-English cross-language information retrieval (CLIR). The main objective of this paper shows that document clusters can effectively resolve the ambiguities tremendously increased in translated queries as well as take into account the context of all the terms in a document. In the framework we propose, a query in Korean/Japanese is first translated into English by looking up bilingual dictionaries, then documents are retrieved for the translated query terms based on the vector space retrieval model or the probabilistic retrieval model. For the top-ranked retrieved documents, query-oriented document clusters are incrementally created and the weight of each retrieved document is re-calculated by using the clusters. In the experiment based on TREC test collection, our method achieved 39.41% and 36.79% improvement for translated queries without ambiguity resolution in Korean-to-English CLIR, and 17.89% and 30.46% improvements in Japanese-to-English CLIR, on the vector space retrieval and on the probabilistic retrieval, respectively. Our method achieved 12.30% improvements for all translation queries, compared with blind feedback in Korean-to-English CLIR. These results indicate that cluster analysis help to resolve ambiguity.

• Korean Journal of Remote Sensing
• /
• v.34 no.5
• /
• pp.721-738
• /
• 2018

The purpose of this study is to create landslide vulnerability using frequency ratio (FR) and evidential belief functions (EBF) model which are two methods of probability model and to select appropriate model for each region through comparison of results in Sacheon-myeon and Jumunjin-eup of Gangneung. 762 locations in Sacheon-myeon and 548 landscapes in Jeonju-eup were constructed based on the interpretation of aerial photographs. Half of each landslide point was randomly selected for modeling and remaining landslides were used for verification purposes. Twenty landslide-inducing factors classified into five categories such as topographic elements, hydrological elements, soil maps (1:5,000), forest maps (1:5,000), and geological maps (1:25,000) were considered for the preparation of landslide vulnerability in the study. The relationship between landslide occurrence and landslide inducing factors was analyzed using FR and EBF models. The two models were then verified using the AUC (curve under area) method. According to the results of verification, the FR model (AUC = 81.2%) was more accurate than the EBF model (AUC = 78.9%) at Jeonjun-eup. In the Sacheon-myeon, the EBF model (AUC = 83.6%) was more accurate than the FR model (AUC = 81.6%). Verification results show that FR model and EBF model have high accuracy with accuracy of around 80%.

• Journal of Navigation and Port Research
• /
• v.47 no.6
• /
• pp.315-323
• /
• 2023

Marine accidents due to loss of stability of small ships have continued to increase over the past decade. In particular, since sudden winds have been pointed out as main causes of most small ship accidents, safety measures have been established to prevent them. In this regard, to prevent accidents caused by sudden winds, a systematic analysis technique is required. The aim of the present study was to develop a probabilistic approach to estimate extreme value and evaluate effects of wind on motion characteristics of ships. The present study included studies of motion analysis, extraction of extreme values, and motion characteristics. A series analysis was conducted for three conditions: wave only, wave with uniform wind speed, and wave with the NPD wind model. Hysteresis filtering and Peak-Valley filtering techniques were applied to time-domain motion analysis results for extreme value extraction. Using extracted extreme values, the goodness of fit test was performed on four distribution functions to select the optimal distribution-function that best expressed extreme values. Motion characteristics of a fishing boat were evaluated for three periodic motion conditions (Heave, Roll, and Pitch) and results were compared. Numerical analysis was performed using a commercial solver, ANSYS-AQWA.

• Journal of Korean Society of Steel Construction
• /
• v.22 no.6
• /
• pp.573-580
• /
• 2010

In this study, an assessment method that considers the effects of directional wind speeds on buildings or structures that are sensitive to wind is proposed. Also, the basic characteristics of directional wind speeds were assessed by means of local annual maximum wind speeds. From the method of assessment of the characteristics of directional wind speeds, their goodness-of-fit was verified by applying extreme value distribution to the data on annual maximum wind speeds from the Korea Meteorological Administration. To consider the characteristics of directional winds, an assessment method is suggested that divides the directional wind pattern of each directional wind speed into four groups. From the study results, all the data on directional wind speeds based on the Gumbel distribution were examined using data on annual maximum wind speeds from Seoul, Tongyung, and Incheon. Since the Gumbel model of all directional wind speeds has independent probability characteristics that govern the 4 directional wind pattern groups, the application ratio proposed was based on the assessment of these four groups. According to the goodness-of-fit of the data on the annual maximum wind speeds based on the Gumbel distribution, new application ratios were proposed that consider the directional wind speeds in Seoul, Tongyung, and Incheon.