• Journal of Korea Multimedia Society
• /
• v.14 no.6
• /
• pp.711-718
• /
• 2011

The k-Nearest Neighbor classifier that does not require a training phase is appropriate for a variable number of classes problem like face recognition, Recently distance metric learning methods that is trained with a given data set have reported the significant improvement of the kNN classifier. However, the performance of a distance metric learning method is variable for each application, In this paper, we focus on the face recognition and compare the performance of the state-of-the-art distance metric learning methods, Our experimental results on the public face databases demonstrate that the Mahalanobis distance metric based on PCA is still competitive with respect to both performance and time complexity in face recognition.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.2
• /
• pp.40-45
• /
• 2000

In speech corpus generation and speech recognition, it is sometimes needed to segment the input speech data without any prior knowledge. A method to accomplish this kind of segmentation, often called as blind segmentation, or acoustic segmentation, is to find boundaries which minimize the Euclidean distances among the feature vectors of each segments. However, the use of this metric alone is prone to errors because of the fluctuations or variations of the feature vectors within a segment. In this paper, we introduce the principal component analysis method to take the trend of feature vectors into consideration, so that the proposed distance measure be the distance between feature vectors and their projected points on the principal components. The proposed distance measure is applied in the LBDP(level building dynamic programming) algorithm for an experimentation of continuous speech segmentation. The result was rather promising, resulting in 3-6% reduction in deletion rate compared to the pure Euclidean measure.

• Journal of KIISE:Computer Systems and Theory
• /
• v.31 no.5_6
• /
• pp.288-296
• /
• 2004

This paper proposes a new mesh simplification algorithm using differential error metric. Many simplification algorithms make use of a distance error metric, but it is hard to measure an accurate geometric error for the high-curvature region even though it has a small distance error measured in distance error metric. This paper proposes a new differential error metric that results in unifying a distance metric and its first and second order differentials, which become tangent vector and curvature metric. Since discrete surfaces may be considered as piecewise linear approximation of unknown smooth surfaces, theses differentials can be estimated and we can construct new concept of differential error metric for discrete surfaces with them. For our simplification algorithm based on iterative edge collapses, this differential error metric can assign the new vertex position maintaining the geometry of an original appearance. In this paper, we clearly show that our simplified results have better quality and smaller geometry error than others.

• Bulletin of the Korean Space Science Society
• /
• 2003.10a
• /
• pp.23-23
• /
• 2003

천체의 거리는 다양한안 방법을 사용하여 직접 혹 간접적으로 측정되어왔다. 특히 우주배경복사를 측정하는 인공위성 관측의 정밀도가 혁명적으로 향상됨에 따라 우주의 나이는 수 % 이내의 정밀도로 결정할 수 있는 수준으로 발전하였다. 우주의 규모가 구체적으로 정의되는 가운데 우주의 절대적인 크기를 제시하기 위하여 천체의 거리를 측정하는데 사용되는 표준등불로 세페이드 변광성 이외에 식쌍성이 새롭게 대두되었다. 이 논문에서는 식쌍성이 거리척도의 표준등불이 될 수 잇는지 검증하기 위하여 광도곡선과 시선속도곡선이 잘 알려진 알골형 쌍성 식쌍성 RY Aqr, RX Gem, RS Vul을 선정하여 거리를 산출하였다. 별을 선정한 기준은 2색 이상의 광도곡선이 발표되고, 이중 분광쌍성으로 시선속도곡선이 각 성분별로 잘 관측되어 발표되고, IUE 관측 자료가 있는 알골형 쌍성이다. 거리 산출과정에서 간접적으로 유추하여 얻는 인자를 줄이기 위하여, 광도곡선으로부터 별의 상대적인 크기를 구하고, 시선속도곡선으로부터 공전궤도의 장반경을 구하고, 별의 에너지 분포 곡선으로부터 별의 온도를 측정하였다. 위 3종류의 관측 결과를 종합하여 식쌍성의 물리적 인자와 거리를 구하였다. 이와 같은 방법으로 구한 거리는 히파크러스를 이용하여 관측한 시차와 비교하였다.

• Review of KIISC
• /
• v.19 no.2
• /
• pp.16-25
• /
• 2009

네트워크 기술의 발달에 따른 서비스의 증가는 네트워크 트래픽과 함께 취약점도 증대하여 이를 악용하는 행위도 늘어나고 있다. 따라서 네트워크 침입탐지 시스템은 증가하는 트래픽의 양을 처리할 수 있어야 하며, 악의적인 행동을 효과적으로 탐지 할 수 있어야 한다. 증가하는 트래픽을 효과적으로 처리하고 탐지의 정확성을 높이기 위해 처리 데이터를 감소시키는 기술이 요구된다. 이러한 방법들은 크게 데이터 필터링, 척도 선택, 데이터 클러스터링의 영역으로 구분되며, 본 논문에서는 척도 선택의 방법으로 데이터 처리의 감소 및 효과적 침입탐지를 수행할 수 있음을 보이고자 한다. 실험 데이터는 KDDCUP 99 데이터 셋을 이용하였으며, 통계적 척도선택의 방법으로 분류율, 오탐율, 거리값, 규칙, 선택된 척도 등을 제시함으로써 침입 탐지 시 데이터 처리량이 감소하였고, 분류율은 증가, 오탐율은 감소하여 침입 탐지 정확성이 높아짐을 알 수 있었다. 또한 본 논문에서 제시한 방법이 다른 관련연구에서 제시한 선택 척도보다 높은 정확성을 보임으로써 보다 유용함을 증명할 수 있었다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.19 no.5
• /
• pp.603-608
• /
• 2009

In this paper, a similarity measure between interval-valued vague sets is proposed. In the interval-valued vague sets representation, the upper bound and the lower bound of a vague set are represented as intervals of interval-valued fuzzy set respectively. Proposed method combines the concept of geometric distance and the center-of-gravity point of interval-valued vague set to evaluate the degree of similarity between interval-valued vague sets. We also prove three properties of the proposed similarity measure. It provides a useful way to measure the degree of similarity between interval-valued vague sets.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.6 s.312
• /
• pp.43-51
• /
• 2006

This paper proposes a metrics for example matching under the example-based machine translation for English-Korean machine translation. Our metrics served as similarity measure is based on edit-distance algorithm, and it is employed to retrieve the most similar example sentences to a given query. Basically it makes use of simple information such as lemma and part-of-speech information of typographically mismatched words. Edit-distance algorithm cannot fully reflect the context of matched word units. In other words, only if matched word units are ordered, it is considered that the contribution of full matching context to similarity is identical to that of partial matching context for the sequence of words in which mismatching word units are intervened. To overcome this drawback, we propose the context-weighting scheme that uses the contiguity information of matched word units to catch the full context. To change the edit-distance metrics representing dissimilarity to similarity metrics, to apply this context-weighted metrics to the example matching problem and also to rank by similarity, we normalize it. In addition, we generalize previous methods using some linguistic information to one representative system. In order to verify the correctness of the proposed context-weighted metrics, we carry out the experiment to compare it with generalized previous methods.

• Journal of KIISE:Software and Applications
• /
• v.36 no.6
• /
• pp.479-490
• /
• 2009

A number of temporal abstraction approaches have been suggested so far to handle the high computational complexity of Markov decision problems (MDPs). Although the structure of temporal abstraction can significantly affect the efficiency of solving the MDP, to our knowledge none of current temporal abstraction approaches explicitly consider the relationship between topology and efficiency. In this paper, we first show that a topological measurement from complex network literature, mean geodesic distance, can reflect the efficiency of solving MDP. Based on this, we build an incremental method to systematically build temporal abstractions using a network model that guarantees a small mean geodesic distance. We test our algorithm on a realistic 3D game environment, and experimental results show that our model has subpolynomial growth of mean geodesic distance according to problem size, which enables efficient solving of resulting MDP.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.20 no.5
• /
• pp.701-709
• /
• 2010

This paper presents an efficient signature recognition method by using the hybrid similarity criterion, which is in inverse proportion to distance and in proportion to correlation between the images. The distance is applied to express the spacial property of image, and the correlation is also applied to express the statistical property. The proposed criterion provides the robust recognition to both the geometrical variations such as position, size, and rotation and the shape variation. The normalized cross-correlation(NCC), which is calculated by considering 4 directions based on the histogram of binary image, is applied to express rapidly and accurately the similarity between the images. The proposed method has been applied to the problem for recognizing the 20 truck images of 288*288 pixels and the 105(3 persons * 35 images) signature images of 256*256 pixels, respectively. The experimental results show that the proposed method has a superior recognition performance that appears the image characters well. Especially, the hybrid criterion of NCC and ordinal distance has a superior recognition performance to the hybrid criterion using city-block or Euclidean distance.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2004.04a
• /
• pp.401-404
• /
• 2004

본 연구에서는 다양한 뉴스그룹들 중에서 사용자의 취향과 유사한 뉴스그룹들을 코호넨 신경망을 이용하여 추천해주는 방법을 제시한다. 신경망을 학습시키기 위한 뉴스 문서의 키워드들을 선택하기 위해 여러 문서들로부터 후보 용어들을 추출하고 퍼지 추론을 적용하여 대표 용어들을 선택한다. 하지만 신경망의 학습패턴을 관찰해 보면, 맡은 부분이 비어있는 희소성 문제를 발견할 수 있다. 이에 본 연구에서는 통계적인 결정계수를 도입하여 불필요한 차원을 제거한 후 신경망을 학습시키는 새로운 방법을 제안한다. 제안된 방법은 모든 차원을 활용할 때 보다 클러스터내 거리와 클러스터간 거리의 척도를 이용한 클러스터 중첩도 면에서 우수한 분류 성능을 보여줌을 확인하였다.