• Spatial Information Research
• /
• v.20 no.4
• /
• pp.47-55
• /
• 2012

Normalized Difference Vegetation Index (NDVI) has been used to measure and monitor plant growth, vegetation cover, and biomass from multispectral satellite data. It is also a valuable index in forest applications, providing forest resource information. In this research, an approach for monitoring forest change using MODIS NDVI time series data is explored. NDVI difference-based approaches for a specific point in time have possible accuracy problems and are lacking in monitoring long-term forest cover change. It means that a multi-time NDVI pattern change needs to be considered. In this study, an efficient methodology to consider long-term NDVI pattern is suggested using a harmonic model. The suggested method reconstructs MODIS NDVI time series data through application of the harmonic model, which corrects missing and erroneous data. Then NDVI pattern is analyzed based on estimated values of the harmonic model. The suggested method was applied to 49 NDVI time series data from Aug. 21, 2009 to Sep. 6, 2011 and its usefulness was shown through an experiment.

• Journal of Digital Contents Society
• /
• v.16 no.5
• /
• pp.671-680
• /
• 2015

The Joint Bayesian[1] method was published in 2012. Since then, it has been used for binary classification in almost all state-of-the-art face recognition methods. However, no improved methods have been published so far except 2D-JB[2]. In this paper we propose an improved version of the JB method that considers the features of both the given face image and its mirror image. In pattern classification, it is very likely to make a mistake when the value of the decision function is close to the decision boundary or the threshold. By making the value of the decision function far from the decision boundary, the proposed method reduces the errors. The experimental results show that the proposed method outperforms the JB and 2D-JB methods by more than 1% in the challenging LFW DB. Many state-of-the-art methods required tons of training data to improve 1% in the LFW DB, but the proposed method can make it in an easy way.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.9
• /
• pp.1299-1305
• /
• 2013

In this paper, we investigate a method of reducing the computational cost in speech recognition by dimensionality reduction of MFCC feature vectors. Eigendecomposition of the feature vectors renders linear transformation of the vectors in such a way that puts the vector components in order of variances. The first component has the largest variance and hence serves as the most important one in relevant pattern classification. Therefore, we might consider a method of reducing the computational cost and achieving no degradation of the recognition performance at the same time by dimensionality reduction through exclusion of the least-variance components. Experimental results show that the MFCC components might be reduced by about half without significant adverse effect on the recognition error rate.

• Journal of Internet Computing and Services
• /
• v.4 no.5
• /
• pp.69-76
• /
• 2003

In this paper we show a new learning algorithm for pattern classification. This algorithm considered a scheme to find a solution to a problem of incremental learning algorithm when the structure becomes too complex by noise patterns included in learning data set. Our approach for this problem uses a pruning method which terminates the learning process with a predefined criterion. In this process, an iterative model with 3 layer feedforward structure is derived from the incremental model by an appropriate manipulations. Notice that this network structure is not full-connected between upper and lower layers. To verify the effectiveness of pruning method, this network is retrained by EBP. From this results, we can find out that the proposed algorithm is effective, as an aspect of a system performence and the node number included in network structure.

• Journal of the Korean Society for Library and Information Science
• /
• v.24
• /
• pp.333-369
• /
• 1993

정보검색 시스템을 대하는 대부분의 이용자의 대답은 '이용하기에 어렵다'라는 것이다. 기계적인 정보검색을 기본 철학으로 하는 기존의 matching paradigm은 정보 곡체를 여기 저기 내용을 옮길 수 있는 물건으로 간주한다. 그리고 기존의 정보시스템은 이용자가 시스템을 구성한 사람의 의도 (즉, indexing, cataloguing rule)를 완전히 이해한다면, 즉 완전하게 질문식(query)을 작성한다면, 효과적인 검색을 할 수 있는 그런 시스템이다. 그러나 어느 이용자가 그 복잡한 시스템을 이해하고 정보검색을 할 수 있겠는가? 한마디로 시스템을 설계한 사람의 의도로 이용자가 적응해서 검색을 한다는 것은 아주 힘든 일이다. 그러나 우리가 이용자에 대한 인식을 다시 한다면 보다 나은 시스템을 만들 수 있다고 본다. 우리 인간은 아주 창조적이어서 자기가 처한 상황에서 이치에 맞게끔 자기 나름대로의 행동을 할 수 있다(sense-making approach). 이 사실을 인식한다면, 왜 이용자들의 행동양식에 시스템 설계자가 적응을 못하는 것인가? 하고 의문을 던질 수 있다. 앞으로의 시스템이 이용자들의 자연스러운 행동 패턴에 맞게 끔 설계된다면 기존의 시스템과 함께 쉽게 이용할 수 있는 편리한 시스템이 설계될 수 있을 것이다. 그러므로 도서관 및 정보학 연구에 있어서 기존의 분류. 목록에 대한 연구와 이용자체에 대한연구(예를 들면, 몇 시에 이용자가 많은가? 어떤 종류의 책을 어떤 계충에서 많이 보는가? 도서 및 잡지가 어떻게 양적으로 성장해 왔는가? 등등의 use study)와 함께 여기서 제시한 제3의 요소인 이용자의 인식(cognition)을 시스템설계에 반드시 도입을 해야만 한다고 본다(user-centric approach). 즉 이용자를 중간 중간에서 도울 수 있는 facilitator가 많이 제공되어야 한다. 이용자의 다양한 패턴의 정보요구(information needs)에 부응할 수 있고, 질문식(query)을 잘 만들 수 없는 이용자를 도울 수 있고(ASK hypothesis: Anomolous State of Knowledge), 어떤 질문식 없이도 자유스럽게 Browsing할 수 있는(예를 들면 hypertext) 시스템을 설계하기 위해서는 눈에 보이는 이용자의 행동패턴(external behavior)도 중요하지만 우리 눈에는 보이지 않는 이용자의 심리상태를 이해한다면 훨씬 나은 시스템을 만들 수 있다. 이용자가 '왜?' '어떤 상황에서,' '어떤 목적으로,' '어떻게,' 정보를 검색하는지에 대해서 새로운 관심을 들려서 이용자들이 얼마나 우리 시스템 설계자들의 의도에 미치지 못한다는 사실을 인식 해야한다. 이 분야의 연구를 위해서는 새로운 paradigm이 필수적으로 필요하다고 본다. 단지 'user-study'만으로는 부족하며 새로운 시각으로 이용자를 연구해야 한다. 가령 새롭게 설치된 computer-assisted system에서 이용자들이 어떻게, 그리핀 어떤 분야에서 왜 그렇게 오류 (error)를 범하는지 분석한다면 앞으로의 computer 시스템 선계에 큰 도움을 줄 수 있을 것으로 믿는다. 실제로 많은 방법이 개발되고 있다. 그러면 시스템 설계자가 가졌던 이용자들이 이러 이러한 방식으로 정보검색을 할 것이라는 예측과(즉, conceptual model) 실제 이용자들이 정보검색을 할 때 일어나는 행동패턴 사이에는(즉, mental model) 상당한 차이점이 있다는 것을 알게 될 것이다. 이 차이점을 줄이는 것이 시스템 설계자의 의무라고 생각한다. 결론적으로, Computer에 대한 새로운 지식과 함께 이용자들의 인식을 연구할 수 있는, 철학적이고 방법론적인 연구를 계속하나가면서, 이용자들의 행동패턴을 어떻게 시스템 설계에 적용할 수 있는 지를 연구해야 한다. 중요하게 인식해야할 사실은 구 Paradigm을 완전히 무시하라는 것은 아니고 단지 이용자에 대한 새로운 인식을 추가하자는 것이다. 그것이 진정한 User Study가 될 수 있는 길이라고 생각하며, 컴퓨터와 이용자 사이의 '원활한 의사교환'이 필수불가결 한 지금 우리 학문이 가야 할 한 연구분야이다. (Human Interaction with Computers)

• The Journal of the Korea Contents Association
• /
• v.7 no.1
• /
• pp.40-47
• /
• 2007

In the protein macro array image, it is important to find out the feature of the each protein chip. A decision error by the personal sense of sight occurred from long time observation while making an experiment in many protein chip image. So the feature extraction is needed by a simulator. In the case of feature analysis for macro array scan image the efficiency is maximized. In the fluorescence scan image, the response for each cell have been depend on R, G, B distribution of color image. But it is difficult to be classified as one color feature in the case of mixed color image. In this paper, the response color of a protein chip is classified according to the fuzzy integral value with respect to fuzzy measure as the user desired color. The result of the experiment for the macro array fluorescence image with the Scan Array 5000 shows that the proposed method using the fuzzy integral is important fact to be make decision for the ambiguous color.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.5
• /
• pp.567-575
• /
• 2020

As online social networks are used as a critical medium for modern people's information sharing and relationship, their users are increasing rapidly every year. This not only increases usage but also surpasses the existing media in terms of information credibility. Therefore, emerging marketing strategies are deliberately attacking social networks. As a result, public opinion, which should be formed naturally, is artificially formed by online attacks, and many people trust it. Therefore, many studies have been conducted to detect agents attacking online social networks. In this paper, we analyze the trends of researches attempting to detect such online social network attackers, focusing on researches using social network graph characteristics. While the existing content-based techniques may represent classification errors due to privacy infringement and changes in attack strategies, the graph-based method proposes a more robust detection method using attacker patterns.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.4
• /
• pp.401-407
• /
• 2009

In this paper, we propose a novel method for the cross-correlation based double-talk detection (DTD), which employing the Gaussian Mixture Model (GMM) in the frequency domain. The proposed algorithm transforms the cross correlation coefficient used in the time domain into 16 channels in the frequency domain using the discrete fourier transform (DFT). The channels are then selected into seven feature vectors for GMM and we identify three different regions such as far-end, double-talk and near-end speech using the likelihood comparison based on those feature vectors. The presented DTD algorithm detects efficiently the double-talk regions without Voice Activity Detector which has been used in conventional cross correlation based double-talk detection. The performance of the proposed algorithm is evaluated under various conditions and yields better results compared with the conventional schemes. especially, show the robustness against detection errors resulting from the background noises or echo path change which one of the key issues in practical DTD.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.8
• /
• pp.501-511
• /
• 2020

This qualitative research examined the experiences and patterns of military life of service maladaptive soldiers. The research site was the place where the soldiers of the A-adaptive soldiers' healing program were conducted, and the study period was conducted from September 2016 to December 2017. Maladaptive soldiers' military life experience pattern was analyzed, and their experience consisted of three domains (early domain after joining the army, middle domain after joining the army, and last domain where they failed to adapt themselves), five cultural themes, 12 categories, and 29 attributes. The cultural themes of maladaptive soldiers in their military life experiences were as follows: facing unfamiliar military culture, hardship, being left alone in a group, pain becoming unbearable, and the last choice of leaving a painful military life. Maladaptive soldiers attempted suicide after they failed to overcome the psychological pain, but they needed help during their early period of adaptation. Moreover, during their middle period of adaptation, they needed guidance for their immature coping strategies, and ultimately they had misperceptions about their death together with a pessimistic view about their life. These results are expected to be used as basic data for the development of mental nursing arbitration programs and suicide prevention projects to help service maladaptive soldiers.

• Journal of Broadcast Engineering
• /
• v.19 no.3
• /
• pp.396-404
• /
• 2014

The field of searching clothing, which is very difficult due to the nature of the informal sector, has been in an effort to reduce the recognition error and computational complexity. However, there is no concrete examples of the whole progress of learning and recognizing for cloth, and the related technologies are still showing many limitations. In this paper, the whole process including identifying both the person and cloth in an image and analyzing both its color and texture pattern is specifically shown for classification. Especially, deformable search descriptor, LBPROT_35 is proposed for identifying the pattern of clothing. The proposed method is scale and rotation invariant, so we can obtain even higher detection rate even though the scale and angle of the image changes. In addition, the color classifier with the color space quantization is proposed not to loose color similarity. In simulation, we build database by training a total of 810 images from the clothing images on the internet, and test some of them. As a result, the proposed method shows a good performance as it has 94.4% matching rate while the former Dense-SIFT method has 63.9%.