• International Journal of Contents
• /
• 제14권2호
• /
• pp.1-6
• /
• 2018

In a wireless mobile communication system, a pilot signal has been considered to be a necessary signal for estimating a changing channel between a base station and a terminal. All mobile communication systems developed so far have a specification for transmitting pilot signals. However, although the pilot signal transmission is easy to estimate the channel,(Ed: unclear wording: it is easy to use the pilot signal transmission to estimate the channel?) it should be minimized because it uses radio resources for data transmission. In this paper, we propose a pilotless channel estimation scheme (PCE) by introducing the clustering method of unsupervised learning used in our deep learning into channel estimation.(Ed: highlight- unclear) The PCE estimates the channel using only the data symbols without using the pilot signal at all. Also, to apply PCE to a real system, we evaluated the performance of PCE based on the resource block (RB), which is a resource allocation unit used in LTE. According to the results of this study, the PCE always provides a better mean square error (MSE) performance than the least square estimator using pilots, although it does not use the pilot signal at all. The MSE performance of the PCE is affected by the number of data symbols used and the frequency selectivity of the channel. In this paper, we provide simulation results considering various effects(Ed: unclear, clarify).

• Journal of Information Science Theory and Practice
• /
• 제6권4호
• /
• pp.28-38
• /
• 2018

Wikipedia is composed of millions of articles, each of which explains a particular entity with various languages in the real world. Since the articles are contributed and edited by a large population of diverse experts with no specific authority, Wikipedia can be seen as a naturally occurring body of human knowledge. In this paper, we propose a method to automatically identify key entities and relations in Wikipedia articles, which can be used for automatic ontology construction. Compared to previous approaches to entity and relation extraction and/or identification from text, our goal is to capture naturally occurring entities and relations from Wikipedia while minimizing artificiality often introduced at the stages of constructing training and testing data. The titles of the articles and anchored phrases in their text are regarded as entities, and their types are automatically classified with minimal training. We attempt to automatically detect and identify possible relations among the entities based on clustering without training data, as opposed to the relation extraction approach that focuses on improvement of accuracy in selecting one of the several target relations for a given pair of entities. While the relation extraction approach with supervised learning requires a significant amount of annotation efforts for a predefined set of relations, our approach attempts to discover relations as they occur naturally. Unlike other unsupervised relation identification work where evaluation of automatically identified relations is done with the correct relations determined a priori by human judges, we attempted to evaluate appropriateness of the naturally occurring clusters of relations involving person-artifact and person-organization entities and their relation names.

• Communications for Statistical Applications and Methods
• /
• 제31권2호
• /
• pp.213-234
• /
• 2024

In recent decades, increasing research attention has been directed toward predicting the price of stocks in financial markets using deep learning methods. For instance, recurrent neural network (RNN) is known to be competitive for datasets with time-series data. Long short term memory (LSTM) further improves RNN by providing an alternative approach to the gradient loss problem. LSTM has its own advantage in predictive accuracy by retaining memory for a longer time. In this paper, we combine both supervised and unsupervised dimension reduction methods with LSTM to enhance the forecasting performance and refer to this as a dimension reduction based LSTM (DR-LSTM) approach. For a supervised dimension reduction method, we use methods such as sliced inverse regression (SIR), sparse SIR, and kernel SIR. Furthermore, principal component analysis (PCA), sparse PCA, and kernel PCA are used as unsupervised dimension reduction methods. Using datasets of real stock market index (S&P 500, STOXX Europe 600, and KOSPI), we present a comparative study on predictive accuracy between six DR-LSTM methods and time series modeling.

• 한국산학기술학회논문지
• /
• 제21권9호
• /
• pp.345-351
• /
• 2020

행정기관은 정책 대응성을 제고하기 위해 빅데이터 분석에 관심을 기울이고 있다. 빅데이터 중 뉴스 기사는 정책 이슈와 정책에 대한 여론을 파악하는데 중요한 자료로 활용될 수 있다. 한편으로 새로운 온라인 매체의 등장으로 뉴스 기사의 생산은 급격히 증가하고 있어 문서 자동분류를 통해 기사를 수집할 필요가 있다. 그러나 기존 뉴스 기사의 범주와 키워드 검색방법으로는 특정 행정기관 및 부서별로 업무에 관련된 기사를 자동적으로 수집하는 것에 한계가 있었다. 또한 기존의 지도학습 기반의 분류 기법은 다량의 학습 데이터가 필요한 단점을 가지고 있다. 이에 본 연구에서는 행정부서의 업무특징을 포함한 분류사전을 활용하여 기사의 분류를 효과적으로 처리하기 위한 방법을 제안한다. 이를 위해 행정 기관의 업무와 신문기사를 Word2Vec와 토픽모델링 기법으로 부서별 특징을 추출하여 분류사전을 생성하고, 행정 부서별로 신문기사를 자동분류 한 결과 71%정도의 정확도를 얻었다. 본 연구는 행정부서별 신문기사를 자동분류하기 위해 부서별 업무 특징 추출 방법과 비지도학습 기반의 자동분류 방법을 제시하였다는 학문적·실무적 기여점이 있다.

• 지능정보연구
• /
• 제21권4호
• /
• pp.1-16
• /
• 2015

교통사고의 원인을 규명하고 미래의 사고를 방지하기 위한 노력의 일환으로 데이터 마이닝 기법을 이용한 교통 데이터 분석의 연구가 이루어지고 있다. 하지만 기존의 교통 데이터를 이용한 마이닝 연구들은 학습된 결과를 사람이 이해하기 어려워 분석에 많은 노력이 필요하다는 문제가 있었다. 본 논문에서는 많은 속성들로 표현된 교통사고 데이터로부터 유용한 패턴을 발견하기 위해 규칙 학습 기반의 데이터 마이닝 기법인 연관규칙 학습기법과 서브그룹 발견기법을 적용하였다. 연관규칙 학습기법은 비지도 학습 기법의 하나로 데이터 내에서 동시에 많이 등장하는 아이템(item)들을 찾아 규칙의 형태로 가공해 주며, 서브그룹 발견기법은 사용자가 지정한 대상 속성이 결론부에 나타나는 규칙을 학습하는 지도학습 기반 기법으로 일반성과 흥미도가 높은 규칙을 학습한다. 규칙 학습 시 사용자의 의도를 반영하기 위해서는 하나 이상의 관심 속성들을 조합한 합성 속성을 만들어 규칙을 학습할 수 있다. 규칙이 도출되고 나면 후처리 과정을 통해 중복된 규칙을 제거하고 유사한 규칙을 일반화하여 규칙들을 더 단순하고 이해하기 쉬운 형태로 가공한다. 교통사고 데이터를 대상으로 두 기법을 적용한 결과 대상 속성을 지정하지 않고 연관규칙 학습기법을 적용하는 경우 사용자가 쉽게 알기 어려운 속성 사이의 숨겨진 관계를 발견할 수 있었으며, 대상 속성을 지정하여 연관규칙 학습기법과 서브그룹 발견기법을 적용하는 경우 파라미터 조정에 많은 노력을 기울여야 하는 연관규칙 학습기법에 비해 서브그룹 발견기법이 흥미로운 규칙들을 더 쉽게 찾을 수 있음을 확인하였다.

• 전자공학회논문지CI
• /
• 제41권2호
• /
• pp.65-76
• /
• 2004

최근 대부분의 문서들이 전자적으로 생성되고 많은 고문서들이 이미지 형태로 전자화되고 있다. 이미지 형태의 전자 문서들은 정보 추출과 데이터베이스화에 많은 어려움이 있기 때문에, 이러한 문서를 효율적으로 관리하고 검색하기 위한 문서구조분석 방법과 문자 인식을 위한 많은 연구가 필요하다. 본 논문은 폰트의 구분 특성(font discrimination features)들이 폰트이미지의 공간적으로 지역적인 특징들에 기반함을 가정한 방법으로써, 객체의 부분기반 표현들을 학습할 수 있는 NMF(non-negative matrix factorization) 알고리즘을 사용하여 폰트를 자동으로 분류하는 방법이다. 제안된 방법은 부분기반의 비지도 학습 방법(part-based unsupervised learning technique)을 이용하여 전체의 폰트 이미지들로부터 각 폰트들의 구분 특징인 부분을 학습하고, 학습된 부분들을 특징으로 사용하여 폰트를 분류하는 방법이다. 실험결과에서 폰트 이미지들의 공간적으로 국부적인 특징들이 조사되고, 그 특징들이 폰트의 식별을 위한 적절성을 보인다. 제안된 방법이 기존의 문자인식, 문서 검색 시스템들의 전처리기로 사용되면, 그 시스템들의 성능을 향상시킬 것으로 기대된다.

• 대한원격탐사학회지
• /
• 제27권6호
• /
• pp.637-651
• /
• 2011

This study presents a rule-based urban image classification method for time series analysis of changes in the vicinity of Asan-si and Cheonan-si in Chungcheongnam-do, using Landsat satellite images (1991-2006). The area has been highly developed through the relocation of industrial facilities, land development, construction of a high-speed railroad, and an extension of the subway. To determine the yearly changing pattern of the urban area, eleven classes were made depending on the trend of development. An algorithm was generalized for the rules to be applied as an unsupervised classification, without the need of training area. The analysis results show that the urban zone of the research area has increased by about 1.53 times, and each correlation graph confirmed the distribution of the Built Up Index (BUI) values for each class. To evaluate the rule-based classification, coverage and accuracy were assessed. When Optimal allowable factor=0.36, the coverage of the rule was 98.4%, and for the test using ground data from 1991 to 2006, overall accuracy was 99.49%. It was confirmed that the method suggested to determine the maximum allowable factor correlates to the accuracy test results using ground data. Among the multiple images, available data was used as best as possible and classification accuracy could be improved since optimal classification to suit objectives was possible. The rule-based urban image classification method is expected to be applied to time series image analyses such as thematic mapping for urban development, urban development, and monitoring of environmental changes.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 1995년도 춘계공동학술대회논문집; 전남대학교; 28-29 Apr. 1995
• /
• pp.587-599
• /
• 1995

This paper proposes a method for pattern recognition using spectral analysis and a hybrid genetic-neural-fuzzy networks. The feature vectors using spectral analysis on contour sequences of 2-D images are extracted, and the vectors are not effected by translation, rotation and scale variance. A combined model using the advantages of conventional method is proposed, those are supervised learning BP, global searching genetic algorithm, and unsupervised learning fuzzy c-method. The proposed method is applied to 10 aircraft recognition to confirm the performance of the method. The experimental results show that the proposed method is better accuracy than conventional method using BP or fuzzy c-method, and learning speed is enhanced.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.273-276
• /
• 1996

We propose a new method to extract valid lines from a visual image. Unsupervised clustering method is used to assign each line to any of the line groups according to its orientation. During the low-level image processing we use an adaptive threshold method to reduce human supervision and to automate the processing sequence. To reduce the misclassification rate and to suppress the superiors line support regions at the clustering stage, the adaptive threshold method is consistently applied. Performing principal component analysis on each line support region provides an efficient method of obtaining line equation. Finally we adopt the theory of robust statistics to guarantee the quality of each extracted line and to eliminate the lines of poor quality. We present the experimental results to verify our method. With the proposed method, one can extract the lines according to the internal orientation similarities and integrate the whole process into one adaptive procedure.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2004년도 Proceedings of ISRS 2004
• /
• pp.389-392
• /
• 2004

Classification of Earth natural components within a full polarimetric SAR image is one of the most important applications of radar polarimetry in remote sensing. In this paper, the unsupervised classification algorithms based on the combined use of the polarimetric processing technique such as the target decomposition and statistical complex Wishart classification method are evaluated and applied to vegetated terrain in Jeju volcanic island.