• Proceedings of the KIEE Conference
• /
• 2011.07a
• /
• pp.1962-1963
• /
• 2011

본 연구에서는 패턴분류를 위해 최적화된 방사형 기저 함수 신경회로망(Radial Basis Function Neural Networks) 분류기를 제안한다. RBFNN은 입력층, 은닉층, 출력층의 3층 구조로 되어 있으며 Multi Dimension, Predictive ability, Robustness한 특징이 있다. RBFNN의 은닉층에는 기존의 활성함수가 아닌 Fuzzy C-means 클러스터링 알고리즘을 사용하여 입력 데이터의 특성을 고려한 적합도를 사용하였다. RBFNN은 은닉층의 노드수와 FCM 클러스터링의 퍼지화 계수, 연결가중치의 다항식 타입이 모델의 성능의 향상에 영향을 미치기 때문에 최적화가 필요하며 본 논문에서는 Differential Evolution(DE) 알고리즘을 사용하여 모델의 구조 및 파라미터를 최적화시켜 모델의 성능을 향상시켰다. 제안된 모델을 평가하기 위해 패턴분류에 많이 사용되는 Iris 데이터와 Wine 데이터를 이용하였다.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.3 no.2
• /
• pp.71-78
• /
• 2008

The sensor node's performance is expected to be improved and the price will be largely decreased. As USN-applied field is expanded, multiple application services are expected to be provided through one USN soon. Multiple applications be required a periodic data reporting, consecutive information monitoring, event-driven data and query-based data as the data pattern that USN is delivered. Accordingly, the mechanism which can assure QoS according to each data characteristic is required. In this paper, A proposed protocol apply the duty cycle of S-MAC protocol variably and flexible according QoS level. This method largely reduces the delay on the delay-sensitive traffic, while keeping the energy efficiency.

• Journal of KIISE:Software and Applications
• /
• v.37 no.2
• /
• pp.120-128
• /
• 2010

A hypernetwork is a generalized hypo-graph and a probabilistic graphical model based on evolutionary learning. Hypernetwork models have been applied to various domains including pattern recognition and bioinformatics. Nevertheless, conventional hypernetwork models have the limitation that they can manage data with categorical or discrete attibutes only since the learning method of hypernetworks is based on equality comparison of hyperedges with learned data. Therefore, real-valued data need to be discretized by preprocessing before learning with hypernetworks. However, discretization causes inevitable information loss and possible decrease of accuracy in pattern classification. To overcome this weakness, we propose a novel feature-wise L1-distance based method for real-valued attributes in learning hypernetwork models in this study. We show that the proposed model improves the classification accuracy compared with conventional hypernetworks and it shows competitive performance over other machine learning methods.

• Proceedings of the Korean Society for Emotion and Sensibility Conference
• /
• 2009.11a
• /
• pp.79-82
• /
• 2009

본 연구는 인터넷의 진화 및 이미징 디바이스 전반의 발달에 따른 디지털 카메라의 진화 전략을 다룬 논문이다. 오늘날의 카메라 기술은 하루가 다르게 발전해 나가고 있다. 수동 필름 카메라에서부터 최첨단 하이브리드(hybrid) 카메라에 이르기 까지 카메라는 끊임없이 진화해 왔으며 사용자 계층 역시 다양해지고 있다. 디지털이 적용된 디지털 카메라의 경우, 웹의 발전과 더불어 활용도가 다양해 짐에 따라서 기존에는 전문가와 비전문가 계층으로 명확하게 나뉘던 사용자 계층도 그 경계가 모호해졌으며 사용자의 수 역시 비약적으로 증가하였다. 수많은 사업자들이 디지털 카메라 시장에 진출함에 따라, 높은 성장세가 지속되던 컴팩트 카메라 (compact camera)시장은 과포화기에 접어들어 성장성이 점차 감소하는 추세에 있으며, 전문가의 영역으로 여겨졌던 DSLR 시장 역시 선발업체의 높은 진입장벽과 이를 극복하기 위한 후발업체의 지속적인 도전으로 인해 경쟁은 치열하고 이익은 내기 어려운 레드 오션(red ocean)이 되었다. 이러한 상황 속에서 디지털 카메라가 활용될 수 있는 대표적 매체인 웹 역시 '웹 2.0' 이라는 용어가 생길 정도로 발전했다. 웹2.0시대가 본격화되어 서로 다른 사용자들이 자유롭게 컨텐츠를 공유하고 누구든지 쉽게 정보의 생산에 동참할 수 있게 됨으로써 기존의 웹에서 일어나지 않았던 다양한 변화들이 생겨나게 되었다. 본 연구에서는 변화의 특징들을 살펴보기 위해 웹 2.0 시대의 대표적인 컨텐츠 공유 커뮤니티로 자리매김한 야후(Yahoo)사(社) 플리커(Flickr)의 사례를 통해 디지털 카메라 사용자들의 디지털 카메라 사용패턴을 살펴보고 이를 통해 디지털 카메라의 진화를 설명하고자 한다. 인터넷이 발전하며 디지털 카메라와 관련 되어 나타난 대표적인 변화는 일반 소비자가 진화하는 점 이라고 볼 수 있다. 컨텐츠의 공유가 자유로운 웹 플랫폼의 등장, 카메라 기술이 결합된 소형 디지털 단말기의 발전 등으로 인해 비전문가 영역에 속해있던 일반 소비자 계층에 Flickr와 같은 웹 커뮤니티에서의 활동을 통해 준전문가 계층으로 진입하려는 욕구가 생기게 되었는데, 이렇게 형성된 욕구에 따라 디지털 카메라 시장 역시 새로운 수요를 만나게 되었다. 본 연구에서는 경쟁이 치열한 시장 상황 속에서 디지털 카메라의 향후 발전 방향을 알아보기 위해 우선 디지털 카메라 기능을 나누어 본 후, 이미징 디바이스 시장과 웹의 발전 형태에 대해 분석해보았다. 이어 본 연구에서는 분류된 디지털 카메라의 기능을 바탕으로 웹 2.0 시대에 시장을 주도할 수 있는 디지털 카메라의 변화방향에 대해서 알아보았다. 본 연구는 감성적 및 이성적 측면에서 디지털 카메라의 향후 전개방향을 살펴보았다는 점에서 이론적 실무적 함의를 가지는 논문이라고 사료된다.

• The KIPS Transactions:PartB
• /
• v.11B no.3
• /
• pp.281-288
• /
• 2004

In order to extract and analyze complex features of the behavior of animals in response to external stimuli such as toxic chemicals, we implemented an adaptive computational method to characterize changes in the behavior of chironomids in response to treatment with the insecticide, diazinon. In this paper, we propose an energy minimization model to extract the features of response behavior of chironomids under toxic treatment, which is applied on the image of velocity vectors. It is based on the improved active contour model and the variations of the energy functional, which are produced by the evolving active contour. The movement tracks of individual chironomid larvae were continuously measured in 0.25 second intervals during the survey period of 4 days before and after the treatment. Velocity on each sample track at 0.25 second intervals was collected in 15-20 minute periods and was subsequently checked to effectively reveal behavioral states of the specimens tested. Active contour was formed around each collection of velocities to gradually evolve to find the optimal boundaries of velocity collections through processes of energy minimization. The active contour which is improved by T. Chan and L. Vese is used in this paper. The energy minimization model effectively revealed characteristic patterns of behavior for the treatment versus no treatment, and identified changes in behavioral states .is the time progressed.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.21 no.6
• /
• pp.749-754
• /
• 2011

In this study, we propose the design of optimized pRBFNNs-based face recognition system using two-dimensional Image and ASM algorithm. usually the existing 2 dimensional face recognition methods have the effects of the scale change of the image, position variation or the backgrounds of an image. In this paper, the face region information obtained from the detected face region is used for the compensation of these defects. In this paper, we use a CCD camera to obtain a picture frame directly. By using histogram equalization method, we can partially enhance the distorted image influenced by natural as well as artificial illumination. AdaBoost algorithm is used for the detection of face image between face and non-face image area. We can butt up personal profile by extracting the both face contour and shape using ASM(Active Shape Model) and then reduce dimension of image data using PCA. The proposed pRBFNNs consists of three functional modules such as the condition part, the conclusion part, and the inference part. In the condition part of fuzzy rules, input space is partitioned with Fuzzy C-Means clustering. In the conclusion part of rules, the connection weight of RBFNNs is represented as three kinds of polynomials such as constant, linear, and quadratic. The essential design parameters (including learning rate, momentum coefficient and fuzzification coefficient) of the networks are optimized by means of Differential Evolution. The proposed pRBFNNs are applied to real-time face image database and then demonstrated from viewpoint of the output performance and recognition rate.

• Journal of Internet Computing and Services
• /
• v.21 no.1
• /
• pp.71-78
• /
• 2020

Recently, as the issue of gentrification emerges, it becomes important to understand the dynamics of local commercial district, which plays the important role for facilitating the local economy and building the community in a city. This paper attempts to provide the framework for systemically analyzing and understanding the local commercial district. Then, this paper empirically analyzes the patterns of sales and flow of floating population by focusing on two representative local commercial districts in Seoul. In addition, the floating population data from telecommunication bases is further modeled with Markov chain for systemically understanding the local commercial districts. Finally, the transition patterns and consumption amounts of floating population are comprehensively analyzed for providing the implications on the evolutions of local commercial districts in a city. We expect that findings of our study could contribute to the economic growth of local commercial district, which could lead to the continuous development of city economy.

• Journal of Korea Spatial Information System Society
• /
• v.11 no.1
• /
• pp.127-136
• /
• 2009

This paper proposes an on-line mining algorithm of moving trajectory patterns in RFID data streams considering changing characteristics over time and constraints of single-pass data scan. Since RFID, sensor, and mobile network technology have been rapidly developed, many researchers have been recently focused on the study of real-time data gathering from real-world and mining the useful patterns from them. Previous researches for sequential patterns or moving trajectory patterns based on stream data have an extremely time-consum ing problem because of multi-pass database scan and tree traversal, and they also did not consider the time-changing characteristics of stream data. The proposed method preserves the sequential strength of 2-lengths frequent patterns in binary relationship table using the time-evolving graph to exactly reflect changes of RFID data stream from time to time. In addition, in order to solve the problem of the repetitive data scans, the proposed algorithm infers candidate k-lengths moving trajectory patterns beforehand at a time point t, and then extracts the patterns after screening the candidate patterns by only one-pass at a time point t+1. Through the experiment, the proposed method shows the superior performance in respect of time and space complexity than the Apriori-like method according as the reduction ratio of candidate sets is about 7 percent.

• Journal of Korea Technology Innovation Society
• /
• v.7 no.3
• /
• pp.607-640
• /
• 2004

This article analyzes the venture firm based upon the new framework of Dynamic Firm Capability (DFC) to identify the process mechanism. Research methodology includes the case study involving structured interview and data collection from two leading Korean ICT(Information Communication Technology) firms in the same sector (DVR). IDIS, spun off from the university, has accumulated the innovative capability based on the R&D department. On the other hand, KODICOM has retained the technological trajectory in terms of marketing competence. Underlying hypothesis is that a firm should show a idiosyncratic evolutionary pattern by acquiring different complimentary assets(CA). In addition, effective internal process should be matched with the essential characteristics not only at the firm level but also at the sectoral level. By analyzing those two different firms, we will find the strategic successful factors based upon the evolutionary point of view. It is a key contribution of this paper to study on the process mechanism of ventures, and to explain detailed process mechanism by viewing two different characteristics of the firm at the functional level.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.5
• /
• pp.375-379
• /
• 2009

Protein domain is the conserved unit of compact tree-dimensional structure and evolution, which carries specific function. Domains may appear in patterns in proteins, since they have been conserved through the evolution for functional formation of proteins. In this paper, we propose a formulated method for conservation analysis of domain combination based on association rule. Proposed method measures mutual dependency of domains in a combination, as well as co-occurrence frequency of them, which is conventionally used. Based on the method, we extracted conserve domain combinations in S.cerevisiae proteins and analyzed their functions based on Gene Ontology. From the results, we drew conclusions that domains in S.cerevisiae proteins form patterns whose members are highly affiliated to one another, and that extracted patterns tend to be associated with molecular function. Moreover, the results testified to proposed method superior to conventional ones for identifying domain combinations conserved for functional cooperation.