• 한국정보통신학회논문지
• /
• 제26권7호
• /
• pp.949-955
• /
• 2022

일반적인 딥러닝 기반의 텍스트 요약 모델은 데이터셋으로부터 자유롭지 않다. 예를 들어 뉴스 데이터셋으로 학습한 요약 모델은 커뮤니티 글, 논문 등의 종류가 다른 글에서 핵심을 제대로 요약해내지 못한다. 본 연구는 이러한 현상을 '데이터 편향 문제'라 정의하고 이를 해결할 수 있는 두 가지 학습 기법을 제안한다. 첫 번째는 고유명사를 마스킹하는 '고유명사 마스킹'이고 두 번째는 텍스트의 길이를 임의로 늘이거나 줄이는 '길이 변화'이다. 또한, 실제 실험을 진행하여 제안 기법이 데이터 편향 문제 해결에 효과적임을 확인하며 향후 발전 방향을 제시한다. 본 연구의 기여는 다음과 같다. 1) 데이터 편향 문제를 정의하고 수치화했다. 2) 요약 데이터의 특징을 바탕으로 학습 기법을 제안하고 실제 실험을 진행했다. 3) 제안 기법은 모든 요약 모델에 적용할 수 있고 구현이 어렵지 않아 실용성이 뛰어나다.

• International Journal of Industrial Entomology and Biomaterials
• /
• 제10권2호
• /
• pp.79-86
• /
• 2005

Molecular markers have increasingly been used in plant genetic analysis, due to their obvious advantages over conventional phenotypic markers, as they are highly polymorphic, more in number, stable across different developmental stages, neutral to selection and least influenced by environmental factors. Among the PCR based marker techniques, ISSR is one of the simplest and widely used techniques, which involves amplification of DNA segment present at an amplifiable distance in between two identical microsatellite repeat regions oriented in opposite direction. Though ISSR markers are dominant like RAPD, they are more stable and reproducible. Because of these properties ISSR markers have recently been found using extensively for finger printing, pohylogenetic analysis, population structure analysis, varietal/line identification, genetic mapping, marker-assisted selection, etc. In mulberry (Morus spp.), ISSR markers were used for analyzing phylogenetic relationship among cultivated varieties, between tropical and temperate mulberry, for solving the vexed problem of identifying taxonomic positions of genotypes, for identifying markers associated with leaf yield attributing characters. As ISSR markers are one of the cheapest and easiest marker systems with high efficiency in generating polymorphism among closely related varieties, they would play a major role in mulberry genome analysis in the future.

• 영재교육연구
• /
• 제5권2호
• /
• pp.17-36
• /
• 1995

The present study examined preschooler's (3-5yrs) representation and evaluation skills in a puzzle completion task. The puzzle contained panels of four children dressed for each seacon and the key to success was using a body scheme to reconstruct the panels (head, torso, legs, feet and sky on top). Baseline data (Study 1) revealed a developmental pattern of increasing bydy scheme representation along with more careful attention to season consitent construction. Spontaneous verbalization also shifted from more guiding statements (where'the head?) to move evaluative statements (this isn't right). Study 2 examined different intervention techniques for increasing representation (verbal laveling) and evaluative processes (error detection practice), along with a control group that had unassisted practice. Three year olds benefited from verbal labeling, four year olds from both types of training. Verbalizations also showed appropriated shifts toward increasing evaluation, particularly for the older children. These findings are discussed in terms of a developmental hypothesis that representation precedes evaluation skills and that training techniques should take into account the relative balance between representation and evaluation skills in the individual for the task at hand.

• 대한조선학회논문집
• /
• 제46권1호
• /
• pp.87-95
• /
• 2009

The applicability of optimization techniques for hull surface fitting has been important in the ship design process. In this research, the Genetic Algorithm has been used as a searching technique for solving surface fitting problem and minimizing errors between B-spline surface and the ship's offset data. The encoded design variables are the location of the vertex points and parametric values. The sufficient accuracy in surface fitting implies not only various techniques for computer-aided design, but also the future production design.

• 한국지능정보시스템학회:학술대회논문집
• /
• 한국지능정보시스템학회 2003년도 춘계학술대회
• /
• pp.474-483
• /
• 2003

A number of studies for corporate bond rating classification problems have demonstrated that artificial intelligence approaches such as Case-based reasoning (CBR) can be alternative methodologies to statistical techniques. CBR is a problem solving technique in that the case specific knowledge of past experience is utilized to find a most similar solution to the new problems. To build a successful CBR system to deal with human information processing, the representation of knowledge of each attribute is an important key factor We propose a hybrid approach of using fuzzy sets that describe the approximate phenomena of the real world because it handles inexact knowledge represented by common linguistic terms in a similar way as human reasoning compared to the other existing techniques. Integration of fuzzy sets with CBR is important to develop effective methods for dealing with vague and incomplete knowledge to statistical represent using membership value of fuzzy sets in CBR.

• Journal of Internet Technology
• /
• 제22권2호
• /
• pp.297-309
• /
• 2021

Mobile sinks are being exploited for various purposes for solving the hotspot problem. Thus, the mobility support technique of mobile sinks is important, and it has been studied steadily. A survey was conducted of studies of mobile sinks in sensor networks having different movement patterns depending on the applications. The related techniques were divided into three main categories according to mobility pattern: predefined, random, and control. In addition, communications for mobile sinks are strongly affected by whether there is a single mobile sink or multiple ones. Based on this two-level categorization, an overview is presented of some state-of-the-art mobility support techniques researched during the past three years, and then the technique are analyzed according to various criteria. Finally, this survey concludes with a summary and a discussion of some future research challenges.

• Asia pacific journal of information systems
• /
• 제16권1호
• /
• pp.71-84
• /
• 2006

Case-based reasoning (CBR) is a reasoning technique that reuses past cases to find a solution to the new problem. It often shows significant promise for improving effectiveness of complex and unstructured decision making. It has been applied to various problem-solving areas including manufacturing, finance and marketing for the reason. However, the design of appropriate case indexing and retrieval mechanisms to improve the performance of CBR is still a challenging issue. Most of the previous studies on CBR have focused on the similarity function or optimization of case features and their weights. According to some of the prior research, however, finding the optimal k parameter for the k-nearest neighbor (k-NN) is also crucial for improving the performance of the CBR system. In spite of the fact, there have been few attempts to optimize the number of neighbors, especially using artificial intelligence (AI) techniques. In this study, we introduce a genetic algorithm (GA) to optimize the number of neighbors to combine. This study applies the novel approach to Korean stock market. Experimental results show that the GA-optimized k-NN approach outperforms other AI techniques for stock market prediction.

• Journal of Electrical Engineering and Technology
• /
• 제10권3호
• /
• pp.944-951
• /
• 2015

Pulse Width Modulation (PWM) techniques are increasingly employed for PWM inverter fed induction motor drive. Among various popular PWM methods used, Selective Harmonic Elimination PWM (SHEPWM) has been widely accepted for its better harmonic elimination capability. In addition, using SHEPWM, it is also possible to maintain better voltage regulation. Hence, in this paper, an attempt has been made to apply Bacterial Foraging Algorithm (BFA) for solving selective harmonic elimination problem. The problem of voltage harmonic elimination together with output voltage regulation is drafted as an optimization task and the solution is sought through proposed method. For performance comparison of BFA, the results obtained are compared with other techniques such as derivative based Newton-Raphson method, and Genetic Algorithm. From the comparison, it can be observed that BFA based approach yields better results. Further, it provides superior convergence, reduced computational burden, and guaranteed global optima. The simulation results are validated through experimental findings.

• 한국지리정보학회지
• /
• 제8권3호
• /
• pp.84-95
• /
• 2005

본 논문은 가시권역의 최대화를 만족하는 가시권 분석에 있어 지형요소가 어떻게 이용될 수 있으며 이러한 최적 다중 가시점 탐색 문제에 있어 지형요소의 이용이 얼마나 효과적인지를 살펴보는 연구이다. 이를 위하여 다양한 지형상태를 반영하는 지역의 DEM 자료와 각 DEM자료에 대한 지형요소 (peak, pass, pit)의 특정을 반영한 여섯 종류의 탐색방법을 제시하고 전통적인 공간 휴리스틱 (spatial heuristic)과의 비교 분석 (계산 시간과 총 가시권역 크기)을 통해서 지형요소를 이용한 방법의 효율성과 적용 가능성을 살펴보았다. 연구결과로써, 가시구역의 중복을 최소화하기 위해 제시된 버퍼링을 이용한 방법의 경우, 비록 공간 휴리스틱 방법에 비해 적은 가시구역 면적을 제시하였지만, 컴퓨팅 시간적인 측면에서 많은 이점을 제공하고 있음을 볼 수 있다. 또한 연구지역의 DEM상의 각각의 개별 그리드 셀을 대상으로 전체 DEM에 대해 계산된 가시구역을 이용한 방법의 경우, 비록 부가적인 계산 시간이 소요됨에도 불구하고 단순한 지형요소를 이용한 방법보다 향상된 분석 결과를 제시하였음을 알 수 있다.

• 산업경영시스템학회지
• /
• 제40권4호
• /
• pp.154-163
• /
• 2017

Collaborative filtering, one of the most widely used techniques to build recommender systems, is based on the idea that users with similar preferences can help one another find useful items. Credit card user behavior analytics show that most customers hold three or less credit cards without duplicates. This behavior is one of the most influential factors to data sparsity. The 'cold-start' problem caused by data sparsity prevents recommender system from providing recommendation properly in the personalized credit card recommendation scenario. We propose a personalized credit card recommender system to address the cold-start problem, using multiple user profiles. The proposed system consists of a training process and an application process using five user profiles. In the training process, the five user profiles are transformed to five user networks based on the cosine similarity, and an integrated user network is derived by weighted sum of each user network. The application process selects k-nearest neighbors (users) from the integrated user network derived in the training process, and recommends three of the most frequently used credit card by the k-nearest neighbors. In order to demonstrate the performance of the proposed system, we conducted experiments with real credit card user data and calculated the F1 Values. The F1 value of the proposed system was compared with that of the existing recommendation techniques. The results show that the proposed system provides better recommendation than the existing techniques. This paper not only contributes to solving the cold start problem that may occur in the personalized credit card recommendation scenario, but also is expected for financial companies to improve customer satisfactions and increase corporate profits by providing recommendation properly.