• 전자공학회논문지
• /
• 제51권7호
• /
• pp.190-200
• /
• 2014

전 세계적으로 방송사 및 영상 관련 기관들의 비디오 기반 시스템이 디지털로 전환되고 있다. 이송 과정에서 발생하는 디지털 드롭아웃은 콘텐츠의 질을 낮추게 만든다. 게다가 디지털 드롭아웃에 초점이 맞춰진 연구가 매우 미미하며 기존 방법들로 해결하기에는 한계점이 존재한다. 상기 이유로, 우리는 디지털 드롭아웃 블록이 가지는 독특한 패턴들의 주파수 특성을 강조할 수 있도록 이산 코사인 변환 (Discrete Cosine Transform) 계수를 기반으로 하는 새로운 특징표현 방법을 제안한다. 또한, 분류를 위해 특징 벡터를 효율적으로 활용할 수 있는 SVM 기반의 오류블록 분류방법을 활용한다. 더 나아가 이 방법은 기존 방법들의 프레임 간 연속성을 이용해 발생하는 문제점들을 극복하였다. 단독 프레임의 정보만을 이용함으로써 빠른 물체의 존재하에서도 동작이 가능하고, 특정 모델이나 추정이 필요하지 않아 최소의 복잡도 하에 오류 검출이 가능하다.

• 대한기계학회논문집A
• /
• 제37권12호
• /
• pp.1445-1452
• /
• 2013

본 논문에서는 단안 카메라를 이용하여 사람과 로봇(카메라)간의 상대위치를 실시간으로 추정하는 알고리즘을 제안한다. HOG(기울기 히스토그램) 특징벡터와 SVM(서포트 벡터 머신) 분류기를 이용하여 사람의 두부 및 어깨영역을 검출한다. 검출된 영역의 크기와 위치를 이용하여 사람과 로봇(카메라)간의 상대 위치 및 각도를 계산한다. 또한 알고리즘 수행속도를 향상시키기 위하여 본 논문에서는 NVIDIA의 GPU와 CUDA 라이브러리를 사용하였다. 그 결과 알고리즘 수행속도는 초당 15 프레임의 영상데이터를 처리할 수 있다. 알고리즘의 정확도 비교를 위해서 SICK 레이저 스캐너 출력과 비교하였다.

• 스마트미디어저널
• /
• 제6권2호
• /
• pp.33-41
• /
• 2017

철도 선로전환기는 열차의 진로를 현재의 궤도에서 다른 궤도로 제어하는 장치이다. 선로전환기의 노후화로 인한 이상 상황은 탈선 등과 같은 심각한 문제를 발생할 수 있기 때문에, 선로전환기의 적절한 교체시기를 결정하는 것은 매우 중요하다. 본 논문에서는 국내 철도 현장에서 획득한 선로전환기의 전류신호로부터 다차원 데이터 큐브를 구성하고 OLAP(On-Line Analytical Processing) 분석을 통하여 체계적으로 "교체가 필요한 데이터"와 "교체 시점이 아닌 데이터" 집합을 정제하여 분류하였다. 또한 선로전환기의 교체시기 탐지 문제를 이진 분류 문제로 해석하여 이진 분류기의 대표적 모델인 SVM(Support Vector Machine)을 탐지기로 설계함으로써 선로전환기의 노후화에 따른 적절한 교체시기를 탐지하는 시스템을 제안한다. 이때, 입력되는 전류 신호를 DWT(Discrete Wavelet Transform)와 PCA(Principal Components Analysis) 기법으로 고차원의 특징벡터 신호를 정보의 손실을 최소화하면서 저차원의 특징벡터로 변환한다. 실제 국내에서 운행 중인 선로전환기의 이상상황 정보가 포함된 대규모의 전류 신호를 이용하여 제안하는 시스템의 성능을 실험적으로 검증한 바 98%를 넘는 탐지 정확도를 확인하였다.

• International Journal of Computer Science & Network Security
• /
• 제23권3호
• /
• pp.67-80
• /
• 2023

Rice is an important food crop for most of the population in Nowadays, psychologists consider social media an important tool to examine mental disorders. Among these disorders, depression is one of the most common yet least cured disease Since abundant of writers having extensive followers express their feelings on social media and depression is significantly increasing, thus, exploring the literary text shared on social media may provide multidimensional features of depressive behaviors: (1) Background: Several studies observed that depressive data contains certain language styles and self-expressing pronouns, but current study provides the evidence that posts appearing with self-expressing pronouns and depressive language styles contain high emotional temperatures. Therefore, the main objective of this study is to examine the literary cyber writers' posts for discovering the symptomatic signs of depression. For this purpose, our research emphases on extracting the data from writers' public social media pages, blogs, and communities; (3) Results: To examine the emotional temperatures and sentences usage between depressive and not depressive groups, we employed the SentiStrength algorithm as a psycholinguistic method, TF-IDF and N-Gram for ranked phrases extraction, and Latent Dirichlet Allocation for topic modelling of the extracted phrases. The results unearth the strong connection between depression and negative emotional temperatures in writer's posts. Moreover, we used Naïve Bayes, Support Vector Machines, Random Forest, and Decision Tree algorithms to validate the classification of depressive and not depressive in terms of sentences, phrases and topics. The results reveal that comparing with others, Support Vectors Machines algorithm validates the classification while attaining highest 79% f-score; (4) Conclusions: Experimental results show that the proposed system outperformed for detection of depression trends in literary cyber writers using sentiment analysis.

• 한국음향학회지
• /
• 제27권3호
• /
• pp.111-118
• /
• 2008

본 논문에서는 음악 장르 분류를 위한 새로운 자동 Taxonomy 구축 알고리즘을 제안한다. 제안된 알고리즘은 모든 가능한 노드들의 분류 확률을 예측하여 예측된 분류 성능값이 가장 좋은 조합을 Taxonomy로 구축하는 것이다. 제안된 알고리즘에서의 분류 확률 예측은 훈련 데이터를 k-fold cross validation을 이용하여 분류기에 적용함으로써 이루어진다. 제안된 알고리즘을 기반으로 한 분류 성능 측정은 2 클래스로 이루어진 각각의 노드에 2개 범주 분류에 효과적인 support vector machine을 적용함으로써 이루어진다. 제안된 알고리즘의 성능 검증을 위해 음색, 리듬, 피치 등 오디오 신호의 특징을 나타내는 다양한 파라미터를 오디오 신호로부터 추출하여 제안된 알고리즘과 기존의 다중 범주 분류기들을 이용하여 분류성능을 평가하였다. 다양한 실험결과 제안된 알고리즘은 기존의 알고리즘에 비하여 5%에서 25%정도의 분류 성능이 향상된 것을 확인할 수 있었고 특히 낮은 차원의 특징벡터를 이용한 분류 실험에서는 10% 에서 25% 향상된 좋은 성능을 보였다.

• IEIE Transactions on Smart Processing and Computing
• /
• 제2권5호
• /
• pp.277-281
• /
• 2013

Modern speaker verification systems based on support vector machines (SVMs) use Gaussian mixture model (GMM) supervectors as their input feature vectors, and the maximum a posteriori (MAP) adaptation is a conventional method for generating speaker-dependent GMMs by adapting a universal background model (UBM). MAP adaptation requires the appropriate amount of input utterance due to the number of model parameters to be estimated. On the other hand, with limited utterances, unreliable MAP adaptation can be performed, which causes adaptation noise even though the Bayesian priors used in the MAP adaptation smooth the movements between the UBM and speaker dependent GMMs. This paper proposes a sparse MAP adaptation method, which is known to perform well in the automatic speech recognition area. By introducing sparse MAP adaptation to the GMM-SVM-based speaker verification system, the adaptation noise can be mitigated effectively. The proposed method utilizes the L0 norm as a regularizer to induce sparsity. The experimental results on the TIMIT database showed that the sparse MAP-based GMM-SVM speaker verification system yields a 42.6% relative reduction in the equal error rate with few additional computations.

• Journal of the korean society of oceanography
• /
• 제32권1호
• /
• pp.1-7
• /
• 1997

Surface current fields in the eastern East China Sea (ECS) were constructed by analyzing trajectories of 58 satellite-tracked surface drifters released during 1991-1996. Composite trajectories and 20-minute-by-20-minute box-averaged current vectors show that the basic current pattern composes of: the Kuroshio main stream, which turns eastward toward the Tokara Strait; a northward branch current of the Kuroshio on the ECS outer shelf deeper than 100 m; and an anticyclonic circulation in the northern Okinawa Trough west of Kyushu. The northward branch current sharply changes its direction to the northeast when it crosses a line connecting Cheju Island, Korea and Goto Islands, Japan. The basic pattern of current field changes slightly from winter to summer, and the main axis of the Tsushima Current in the Korea Strait is found to shift seasonally. The drifter experiment does not support the claim that the Yellow Sea Warm Current is separated from the northward branch current on the outer shelf southeast of Cheju Island. We suggest that the use of the term 'Tsushima Current' be limited to the northeast channel flow in the Korea Strait. The new term 'Kuroshio Branch Current' is suggested for the northward branch current on the outer shelf south of Cheju-do, which is separated from the Kuroshio.

• Journal of Electrical Engineering and Technology
• /
• 제12권6호
• /
• pp.2118-2126
• /
• 2017

The large-scale power system blackouts have indicated that conventional protection relays that based on local signals cannot fit for modern power grids with complicated setting or heavily loaded-flow transfer. In order to accurately detect various faulted lines and improve the fault-tolerance of wide-area protection, a novel multi-section weighed fault matching and detecting algorithm is proposed. The real protection vector (RPV) and expected section protection vectors (ESPVs) for five fault sections are constructed respectively. The function of multi-section weighed fault matching is established to calculate the section fault matching degrees between RPV and five ESPVs. Then the fault degree of protected line based on five section fault degrees can be obtained. Two fault detecting criterions are given to support the higher accuracy rate of detecting fault. With the enumerating method, the simulation tests illustrate the correctness and fault-tolerance of proposed algorithm. It can reach the target of 100% accuracy rate under 5 bits error of wide-area protections. The influence factors of fault-tolerance are analyzed, which include the choosing of wide-area protections, as well as the topological structures of power grid and fault threshold.

• 한국생물정보학회:학술대회논문집
• /
• 한국생물정보시스템생물학회 2003년도 제2차 연례학술대회 발표논문집
• /
• pp.133-138
• /
• 2003

The prediction of protein secondary structure has been an important bioinformatics tool that is an essential component of the template-based protein tertiary structure prediction process. It has been known that the predicted secondary structure information improves both the fold recognition performance and the alignment accuracy. In this paper, we describe several novel ideas that may improve the prediction accuracy. The main idea is motivated by an observation that the protein's structural information, especially when it is combined with the evolutionary information, significantly improves the accuracy of the predicted tertiary structure. From the non-redundant set of protein structures, we derive the 'potential' parameters for the protein secondary structure prediction that contains the structural information of proteins, by following the procedure similar to the way to derive the directional information table of GOR method. Those potential parameters are combined with the frequency matrices obtained by running PSI-BLAST to construct the feature vectors that are used to train the support vector machines (SVM) to build the secondary structure classifiers. Moreover, the problem of huge model file size, which is one of the known shortcomings of SVM, is partially overcome by reducing the size of training data by filtering out the redundancy not only at the protein level but also at the feature vector level. A preliminary result measured by the average three-state prediction accuracy is encouraging.

• 한국멀티미디어학회논문지
• /
• 제18권10호
• /
• pp.1197-1204
• /
• 2015

Smoke detection plays an important role for the early detection of fire. In this paper, we suggest a newly developed method that generated LBPV(Local Binary Pattern Variance)s as special feature vectors from RGB contrast images can be applied to detect smoke using SVM(Support Vector Machine). The proposed method rearranges mean value of the block from each R, G, B channel and its intensity of the mean value. Additionally, it generates RGB contrast image which indicates each RGB channel’s contrast via smoke’s achromatic color. Uniform LBPV, Rotation-Invariance LBPV, Rotation-Invariance Uniform LBPV are applied to RGB Contrast images so that it could generate feature vector from the form of LBP. It helps to distinguish between smoke and non smoke area through SVM. Experimental results show that true positive detection rate is similar but false positive detection rate has been improved, although the proposed method reduced numbers of feature vector in half comparing with the existing method with LBP and LBPV.