• Annual Conference on Human and Language Technology
• /
• 1999.10e
• /
• pp.209-214
• /
• 1999

패턴기반기계번역방식은 원시언어패턴과 그에 대한 대역언어패턴들의 쌍을 이용하여 구문분석과 변환을 수행하는 기계번역방식이다. 패턴기반 기계번역방식은 번역할 때 발생하는 애매성을 해소하기 위해 패턴의 길이를 문장단위까지 늘이기 때문에, 패턴의 수가 급증하는 문제점을 가진다. 본 논문에서는 패턴의 단위를 구단위로 한정시킬 때 발생하는 애매성을 해소하는 방법으로 시소러스를 기반으로 한 두단계 대역어 선택 방식을 제안함으로써 효과적으로 애매성을 감소시키면서 패턴의 길이를 줄이는 모델을 제시한다. 두단계 대역어 선택 방식은 원시언어의 한 패턴에 대해 여러 가능한 목적언어의 대역패턴들이 있을 때, 첫 번째 단계에서는 원시언어 내에서의 제약조건에 맞는 몇가지 대역패턴들을 선택하고, 두번째 단계에서는 목적언어 내에서의 제약조건에 가장 적합한 하나의 대역패턴을 선택하는 방식이다. 또한 본 논문에서는 이와 같은 모델에서 패턴의 수가 코퍼스의 증가에 따른 수렴가능성을 논한다.

• Journal of KIISE:Software and Applications
• /
• v.27 no.1
• /
• pp.41-49
• /
• 2000

This paper describes a method of generating MPHFs(Minimal Perfect Hash Functions) for large static search key sets. The MOS(Mapping-Ordering-Searching) approach is widely used presently in MPHF generation. In this research, the MOS approach is improved and a SOMS(Selecting-Ordering-Mapping-Searching) approach is proposed, where the Selecting step is newly introduced and the Orderng step is performed before the Mapping step to generate MPHFs more effectively. The MPHF generation algorithm proposed in this research is probabilistic and the expected processing time is linear to the number of keys. Experimental results show that MPHFs are generated fast and the space needed to represent the hash functions is small.

• Journal of rehabilitation welfare engineering & assistive technology
• /
• v.7 no.2
• /
• pp.19-26
• /
• 2013

This paper propose a surface EMG signal based gait phase recognition method that selects features and channels adaptively. The proposed method can be used to control powered artificial prosthetic for lower limb amputees and can reduce overhead in real-time pattern recognition by selecting adaptive channels and features in an embedded device. The method can enhance the classification accuracy by adaptively selecting channels and features based on sensitivity and specificity of each subject because EMG signal patterns may vary according to subject's locomotion convention. In the experiments, we found that the muscles with highest recognition rate are different between human subjects. The results also show that the average accuracy of the proposed method is about 91% whereas those of existing methods using all channels and/or features is about 50%. Therefore we assure that sEMG signal based gait phase recognition using small number of adaptive muscles and corresponding features can be applied to control powered artificial prosthetic for lower limb amputees.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.13 no.4
• /
• pp.107-116
• /
• 2013

This algorithm suggests a method in which a minimum spanning tree can be obtained fast by reducing the number of an algorithm execution. The suggested algorithm performs a select-and-delete process. In the select process, firstly, it performs Borůvka's first stage for all the vertices of a graph. Then it re-performs Borůvka's first stage for specific vertices and reduces the population of the edges. In the delete process, it deletes the maximum weight edge if any cycle occurs between the 3 edges of the edges with the reduced population. After, among the remaining edges, applying the valency concept, it gets rid of maximum weight edges. Finally, it eliminates the maximum weight edges if a cycle happens among the vertices with a big valency. The select-and-delete algorithm was applied to 9 various graphs and was evaluated its applicability. The suggested select process is believed to be the vest way to choose the edges, since it showed that it chose less number of big edges from 6 graphs, and only from 3 graphs, comparing to the number of edges that is to be performed when using MST algorithm. When applied the delete process to Kruskal algorithm, the number of performances by Kruskal was less in 6 graphs, but 1 more in each 3 graph. Also, when using the suggested delete process, 1 graph performed only the 1st stage, 5 graphs till 2nd stage, and the remaining till 3rd stage. Finally, the select-and-delete algorithm showed its least number of performances among the MST algorithms.

• Journal of the Korea Society for Simulation
• /
• v.23 no.4
• /
• pp.189-195
• /
• 2014

In this paper, we conducted a study on place-free and time-free cloud computing (CC) adoption decision-making model. Panel survey data which is collected from 65 people and CART (classification and regression tree) which is one of data mining approaches are used to construct decision-making model. In this modeling, there are 2 steps: In the first step, significant questions (variables) are selected. After that, the CART decision-making model is constructed using the selected variables. In the variable selection stage, the 25 questions are reduced to 5 ones. The benefits of question reduction are quick response from respondent and reducing model-construction time.

• Journal of the Korean Society for Railway
• /
• v.19 no.5
• /
• pp.663-671
• /
• 2016

This paper investigates how to change passengers' preferences for express trains according to line extension in Seoul Metro Line 9. Before-and-after line extension, we traced passengers' path choices using Smart Card data via the method suggested by Kim et al. (2015). We developed two multinomial logit models before-and-after line extension. Transferability test showed that there is no difference between the two models. However, the load factor of the express train increased by 6.7% and the gap between the local and express trains became significantly wider.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.06b
• /
• pp.139-141
• /
• 2006

유전자 알고리즘에서 좋은 염색체(chromosome)를 선택하는 방법은 알고리즘의 성능을 향상시키는데 매우 중요한 핵심 요소이다. 이러한 선택 기법 중에는 비례 선택 기법, 순위기반 선택 기법, 토너먼트 선택기법 등이 잘 알려져 있다. 이 중 가장 성능이 좋은 토너먼트 선택 기법은 열성 염색체중 우성인 유전자를 포함하는 열성 염색체가 선택에서 배제되어 지역적 최적해(local minima)를 구할 가능성, 열성 염색체가 다음 세대 진화를 방해할 가능성 등의 문제점을 가지고 있다. 본 논문에서는 토너먼트 선택 기법의 문제점을 해결하기 위해서 토너먼트-교배 선택 기법을 제안하였다. 이 방법은 토너먼트 선택 기법을 기반으로 하되 열성 염색체가 선택되었을 경우 그 안에 들어 있는 우성 유전자를 알고리즘 진화에 반영시키고자 교배 단계를 추가한 기법이다. 제안된 토너먼트-교배 선택 기법을 이용하면 기존의 토너먼트 선택 기법보다 평균수행시간이 짧아져 해에 수렴하는 속도가 향상된다.

• Proceedings of the Korea Contents Association Conference
• /
• 2011.05a
• /
• pp.99-100
• /
• 2011

21세기에 들어서면서 급속도로 성장한 한국의 게임 산업은 어느덧 국내에서 성숙기 단계를 맞이하고 있다. 또한 게임 산업의 꾸준한 확장을 위해서는 글로벌 시장에서 인정받을 수 있는 게임 개발이 절실히 요구되고 있다. 본 논문은 이를 위한 한 단계로서 MMORPG 장르의 게임 중에 오랫동안 이 분야에서 성공 가도를 달리고 있는 월드오브워크래프트를 몰입도 관점에서 성공 요인을 분석하고, 이를 바탕으로 2011년도에 주목 받고 있는 신생 게임 테라의 특징을 동일한 관점에서 분석한다. 몰입도 분석 방법은 정보이론과 정신물리학에 기반하며 게임의 레벨과 스킬의 개수에 의존적인 함수로서 정량적인 접근 방법을 제안한다. 제안 방법에 따르면 월드오브워크래프트는 확장팩 대격변을 전후하여 레벨에 따른 스킬 콘텐츠를 재배치하여 새로운 시도를 선보였다. 그 반면에 테라는 화려한 그래픽을 강점으로 내세우지만 낮은 몰입도와 레벨에 따른 스킬 콘텐츠가 부족하고 그 의도마저 모호하다. 이러한 특징은 실제 테라를 플레이할 때 나타나는데 월드오브워크래프트에 비해 상대적으로 단조로운 스킬 선택권과 직업 선택권의 형태로 발견되어지는 문제로서 제안하는 방법이 몰입도를 측정하는 하나의 방법으로 평가된다.

• The Korean Journal of Applied Statistics
• /
• v.35 no.5
• /
• pp.667-684
• /
• 2022

Gene expression data present the level of mRNA abundance of each gene, and analyses of gene expressions have provided key ideas for understanding the mechanism of diseases and developing new drugs and therapies. Nowadays high-throughput technologies such as DNA microarray and RNA-sequencing enabled the simultaneous measurement of thousands of gene expressions, giving rise to a characteristic of gene expression data known as high dimensionality. Due to the high-dimensionality, learning models to analyze gene expression data are prone to overfitting problems, and to solve this issue, dimension reduction or feature selection techniques are commonly used as a preprocessing step. In particular, we can remove irrelevant and redundant genes and identify important genes using gene selection methods in the preprocessing step. Various gene selection methods have been developed in the context of machine learning so far. In this paper, we intensively review recent works on gene selection methods using machine learning approaches. In addition, the underlying difficulties with current gene selection methods as well as future research directions are discussed.

• Journal of KIISE:Software and Applications
• /
• v.30 no.3_4
• /
• pp.278-284
• /
• 2003

Speech act is the speaker's intentions indicated through utterances. It is important for understanding natural language dialogues and generating responses. This paper proposes the method of two stage that increases the performance of the korean speech act decision. The first stage is to select features from the part of speech results in sentence and from the context that uses previous speech acts. We use x$^2$ statistics(CHI) for selecting features that have showed high performance in text categorization. The second stage is to determine speech act with selected features and Neural Network. The proposed method shows the possibility of automatic speech act decision using only POS results, makes good performance by using the higher informative features and speed up by decreasing the number of features. We tested the system using our proposed method in Korean dialogue corpus transcribed from recording in real fields, and this corpus consists of 10,285 utterances and 17 speech acts. We trained it with 8,349 utterances and have test it with 1,936 utterances, obtained the correct speech act for 1,709 utterances(88.3%). This result is about 8% higher accuracy than without selecting features.