• Title/Summary/Keyword: network clustering algorithm

Search Result 536, Processing Time 0.029 seconds

A Comparative Study on Image Enhancement Methods for Low Contrast Images (저대비 영상을 위한 영상향상 기법들의 비교연구)

  • Kim Yong-Soo;Kim Nam-Jin;Lee Se-Yul
    • Proceedings of the Korean Institute of Intelligent Systems Conference
    • /
    • 2005.04a
    • /
    • pp.269-272
    • /
    • 2005
  • The principal objective of enhancement methods is to process an image so that the result is more suitable than the original image for a specific application. Images taken in the night can be low-contrast images because of poor environments. In this paper, we compare the structure of ICECA(Image Contrast Enhancement technique using Clustering Algorithm) with the structures of HE(Histogram Equalization), BBHE(Brightness preserving Bi-Histogram Equalization), and Multi -Scale Retinex(MSR). We compared performances of image enhancement methods by applying these methods to a set of diverse images.

  • PDF

Rule extraction from trained neural network using NofM algorithm with improved clustering step (개선된 군집화 단계의 NofM 알고리즘을 이용한 훈련된 신경망으로부터의 규칙추출)

  • Lee, Han-Yul;Ra, Jong-Hei;Kim, Moon-Hyun
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2001.10a
    • /
    • pp.581-584
    • /
    • 2001
  • 신경망이 만들어내는 출력에 대한 정보는 수치적으로 분산되어 신경망에 저장되므로, 인간이 직접 해석하기가 힘들다. 본 논문에서는 LRE(link rule extraction)기법인 NofM 알고리즘의 6단계 중에서 초기 단계인 가중치 군집화 단계를 개선하여 추출되는 규칙들의 전제부에 들어가는 규칙 조건들의 수를 조절함으로써, 추출된 규칙이 입력 특성에 대한 정보를 과잉 일반화하거나, 과잉 구체화하는 것을 피할 수 있음을 실험을 통해 보였다. 일반적으로 NofM 알고리즘에서 가중치들을 군집화한 때는 Join 알고리즘을 사용하는데, 본 논문에서는 Join 알고리즘의 Join condition을 0.05부터 0.25까지 0.05씩 점진적으로 확대하여 클러스터링을 하여줌으로써 신경망의 출력에 중요한 역할을 하는 가중치들을 효과적으로 군집화함을 보였다.

  • PDF

Semantic Network Automatic Clustering Method of the Unified Medical Language System Using Genetic Algorithm (유전자 알고리즘을 이용한 통합의학언어시스템(UMLS)의 의미망 자동 군집 방법)

  • 지영신;김태준;전혜경;정헌만;이정현
    • Proceedings of the Korean Information Science Society Conference
    • /
    • 2003.10a
    • /
    • pp.82-84
    • /
    • 2003
  • UMLS 의미망은 크기가 방대하고 복잡하여 사용자가 이해하기가 어렵고 화면상에 모든 의미망을 모두 표현할 수 없다는 단점을 가지고 있다. 이 문제를 해결하기 위해 의미망을 효율적으로 분할하기 위한 규칙들이 소개되고 있지만 이것은 UMLS 의미망이 수정될 때마다 규칙을 적용하여 수작업으로 분류를 해야한다는 단점이 있다. 이 문제점을 해결하기 위해 유전자 알고리즘을 이용한 UMLS 의미망의 자동 군집화 방법을 제안한다. 제안한 방법은 각각의 의미유형 간의 연결된 의미관계를 사용하여 의미망을 구조적으로 유사한 의미유형 집합들로 군집화하고 규칙에 의한 군집 방법의 결과 비교 평가한다.

  • PDF

Estimation of pattern classification vigilance parameter using neural network

  • Son, Jun-Hyug;Seo, Bo-Hyeok
    • Proceedings of the KIEE Conference
    • /
    • 2004.05a
    • /
    • pp.95-97
    • /
    • 2004
  • This paper estimates Adaptive Resonance Theory 1(ART1) as a vigilance parameter of pattern clustering algorithm. Inherent characteristics of the model are analyzed. In particular the vigilance parameter ${\rho}$ and its role in classification of patterns is examined. Our estimates show that the vigilance parameter as designed originally does not necessarily increase the number of categories with its value but can decrease also. This is against the claim of solving the stability-plasticity dilemma. However, we have proposed a modified vigilance parameter estimate criterion which takes into account the problem of subset and superset patterns and stably categorizes arbitrarily many input patterns in one list presentation when the vigilance parameter is closer to one.

  • PDF

Find Friends System on SNS to Apply Clustering Algorithm in Network Environment (클러스터링 알고리즘의 네트워크 환경 적용을 통한 SNS 친구추천)

  • Lee, Rich C.;Lee, Woo-Key;Park, Simon S.
    • Proceedings of the Korean Information Science Society Conference
    • /
    • 2012.06c
    • /
    • pp.31-32
    • /
    • 2012
  • 본 연구는 소셜 네트워크에서 사용자간의 친밀도에 기반하여 보여주는 '친구추천' 이라는 방법을 그래프 클러스터링을 이용하여 접근하고자 한다. 기존의 방법과는 다르게 사용자에게 개인화된 선별 정보를 제공하는데 목적이 있다. 또한 일반적 클러스터링이 아닌 그래프 이론에 근거한 거리 계산을 기반으로 친화력 전파 모델(Affinity Propagation) 클러스터링 기법을 적용하는 방법을 제안한다. 이 방법으로 클러스터링을 진행하여 선별된 같은 그룹 안에 있는 개인화된 친구 추천을 효과적으로 수행할 수 있음을 입증하였다.

Effective Mobile Data Offloading using DBSCAN (DBSCAN을 사용한 효과적인 모바일 데이터 오프로딩)

  • Kim, SeungKeun;Yang, Sung-Bong
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2018.05a
    • /
    • pp.81-84
    • /
    • 2018
  • Recently, many researchers claim that mobile data offloading is a key solution to alleviating overloaded cellular traffic by dividing the overloaded traffic with femtocells, WiFi networks or users. In this paper, we propose an idea to select a group of users, known as VIPs, that is able to effectively transfer the data to others using Density-Based Spatial Clustering of Application with Noise, also known as DBSCAN algorithm. We conducted our experiments using NCCU real trace dataset. The results show that our proposed idea offload about 70~77% of the network with VIP set size of four, which is better than the compared methods.

HAPS Network MBS placement with EM Clustering Algorithm (HAPS 기반 네트워크에서의 실시간 이동 기지국 위치 문제 해결 정책)

  • Woong-Hee Jung;Ha Yoon Song;Kwan Sik Cho
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2008.11a
    • /
    • pp.1307-1310
    • /
    • 2008
  • EM(Expectation Maximization)은 불확실한 데이터들을 가지고 분포를 모델링하는, 널리 알려진 군집화 알고리즘이다. EM 알고리즘에서, 정규 분포는 기대(Expectation)-최대화(Maximization)과정을 반복하는 과정에서 그 윤곽을 다져간다. 이 때 이 과정은 EM 알고리즘의 다양한 확률 초기화에 따라 다른 결과를 내게 된다, 본 논문에서는 이 확률 초기화 값의 조정을 통하여 HAPS(High Altitude Platform Station) 기반 네트워크에서 이동 기지국의 위치를 실시간으로 결정하고자 하는 문제를 풀기 위한 조건을 몇 가지 반영시켜 확률 초기 값을 결정해 보고, 그 결과를 제시한다. 이에 더불어, ITU에서 제한하고 있는 이동 기지국의 서비스 반경을 고려하는 방법을 제시한다.

The Algorithm for an Energy-efficient Particle Sensor Applied LEACH Routing Protocol in Wireless Sensor Networks (무선센서네트워크에서 LEACH 라우팅 프로토콜을 적용한 파티클 센서의 에너지 효율적인 알고리즘)

  • Hong, Sung-Hwa;Kim, Hoon-Ki
    • Journal of the Korea Society for Simulation
    • /
    • v.18 no.3
    • /
    • pp.13-21
    • /
    • 2009
  • The sensor nodes that form a wireless sensor network must perform both routing and sensing roles, since each sensor node always has a regular energy drain. The majority of sensors being used in wireless sensor networks are either unmanned or operated in environments that make them difficult for humans to approach. Furthermore, since many wireless sensor networks contain large numbers of sensors, thus requiring the sensor nodes to be small in size and cheap in price, the amount of power that can be supplied to the nodes and their data processing capacity are both limited. In this paper, we proposes the WSN(Wireless Sensor Network) algorithm which is applied sensor node that has low power consumption and efficiency measurement. Moreover, the efficiency routing protocol is proposed in this paper. The proposed algorithm reduces power consumption of sensor node data communication. It has not researched in LEACH(Low-Energy Adaptive Clustering Hierarchy) routing protocol. As controlling the active/sleep mode based on the measured data by sensor node, the energy consumption is able to be managed. In the event, the data is transferred to the local cluster head already set. The other side, this algorithm send the data as dependent on the information such as initial and present energy, and the number of rounds that are transformed into cluster header and then transferred. In this situation, the assignment of each node to cluster head evenly is very important. We selected cluster head efficiently and uniformly distributed the energy to each cluster node through the proposed algorithm. Consequently, this caused the extension of the WSN life time.

The Compression of Normal Vectors to Prevent Visulal Distortion in Shading 3D Mesh Models (3D 메쉬 모델의 쉐이딩 시 시각적 왜곡을 방지하는 법선 벡터 압축에 관한 연구)

  • Mun, Hyun-Sik;Jeong, Chae-Bong;Kim, Jay-Jung
    • Korean Journal of Computational Design and Engineering
    • /
    • v.13 no.1
    • /
    • pp.1-7
    • /
    • 2008
  • Data compression becomes increasingly an important issue for reducing data storage spaces as well as transmis-sion time in network environments. In 3D geometric models, the normal vectors of faces or meshes take a major portion of the data so that the compression of the vectors, which involves the trade off between the distortion of the images and compression ratios, plays a key role in reducing the size of the models. So, raising the compression ratio when the normal vector is compressed and minimizing the visual distortion of shape model's shading after compression are important. According to the recent papers, normal vector compression is useful to heighten com-pression ratio and to improve memory efficiency. But, the study about distortion of shading when the normal vector is compressed is rare relatively. In this paper, new normal vector compression method which is clustering normal vectors and assigning Representative Normal Vector (RNV) to each cluster and using the angular deviation from actual normal vector is proposed. And, using this new method, Visually Undistinguishable Lossy Compression (VULC) algorithm which distortion of shape model's shading by angular deviation of normal vector cannot be identified visually has been developed. And, being applied to the complicated shape models, this algorithm gave a good effectiveness.

Energy Efficient Cluster Head Election Algorithm Considering RF-Coverage (RF-Coverage를 고려한 에너지 효율적인 클러스터 헤드 선출 알고리즘)

  • Lee, Doo-Wan;Han, Youn-Hee;Jang, Kyung-Sik
    • Journal of the Korea Institute of Information and Communication Engineering
    • /
    • v.15 no.4
    • /
    • pp.993-999
    • /
    • 2011
  • In WSN, at the initial stage, sensor nodes are randomly deployed over the region of interest, and self-configure the clustered networks by grouping a bunch of sensor nodes and selecting a cluster header among them. Specially, in WSN environment, in which the administrator's intervention is restricted, the self-configuration capability is essential to establish a power-conservative WSN which provides broad sensing coverage and communication coverage. In this paper, we propose a communication coverage-aware cluster head election algorithm for Herearchical WSNs which consists of communication coverage-aware of the Base station is the cluster head node is elected and a clustering.