• Proceedings of the Korea Information Processing Society Conference
• /
• 2011.04a
• /
• pp.1245-1246
• /
• 2011

최근 다양한 논문 검색 사이트들이 등장함에 따라 논문 데이터에 대한 관심이 높아지고 있다. 본 논문에서는 논문 데이터 연결망을 다양한 실험을 통해 분석한다. 더 나아가, 기존의 웹 연결망, 블로그 연결망과의 특성 비교를 통해 연결망 간의 차이를 보이고, 이러한 차이가 나타나는 원인을 규명한다.

• Journal of KIISE:Computer Systems and Theory
• /
• v.35 no.9_10
• /
• pp.421-428
• /
• 2008

A. Ghafoor proposed Even networks as a class of fault-tolerant multiprocessor networks in [1] and analyzed so many useful properties include node disjoint paths. By introducing node disjoint paths in [1], fault diameter of Even networks is d+2(d=odd) and d+3(d=even). But the lengths of node disjoint paths proved in [1] are not the shortest. In this paper, we show that Even network Ed is node symmetric. We also propose the shortest lengths of node disjoint paths using cyclic permutation, and fault diameter of Even networks is d+1.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.10a
• /
• pp.58-60
• /
• 2004

최근에 하이퍼큐브의 망비용을 개선한 하이퍼-스타 연결망이 제안되었다. 본 논문에서는 하이퍼-스타 연결망 HS(2n, n)의 container를 이용하여 k-wide diameter가 dist(u, v)＋4이하임과 HS(2n, n)의 고장지름이 D(HS(2n, n))＋2 이하임을 보인다.

• Korean Journal of Environment and Ecology
• /
• v.25 no.3
• /
• pp.412-420
• /
• 2011

This study was conducted to quantitatively analyze the temporal change of green space network at multi-scales from 1975 to 2006 with the application of graph theory in Seoul, Korea. Remarkable change of connectivity was detected in green space networks at the scale ranging from 1,000 ~ 1,600 m during 30 years. Green spaces and their networks have been restoring after 1990 since forest areas had been fragmented in 1975. In 2006, we identified the important core habitat areas that can sustain diverse wildlife species and stepping stones composed of small patches that can link these core habitat areas. Green spaces showed high correlation with the relative importance value of green space connectivity. So, this study could graphically represent green space networks of Seoul City. Green spaces of core areas distributed at the northern and southern boundary, and those of stepping stones possessing the high value of betweenness centrality consisted at the middle, eastern and western boundary. These results indicate that green space network can be graphically and quantitatively explained by degree centrality, betweenness centrality and relative importance value of connectivity with the application of graph theory.

• Sleep Medicine and Psychophysiology
• /
• v.25 no.2
• /
• pp.82-91
• /
• 2018

Objectives: Synchronous electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI) has been used to explore sleep stage dependent functional brain networks. Despite a growing number of sleep studies using EEG-fMRI, few studies have conducted network analysis on whole night sleep due to difficulty in data acquisition, artifacts, and sleep management within the MRI scanner. Methods: In order to perform network analysis for whole night sleep, we proposed experimental procedures and data processing techniques for EEG-fMRI. We acquired 6-7 hours of EEG-fMRI data per participant and conducted signal processing to reduce artifacts in both EEG and fMRI. We then generated a functional brain atlas with 68 brain regions using independent component analysis of sleep fMRI data. Using this functional atlas, we constructed sleep level dependent functional brain networks. Results: When we evaluated functional connectivity distribution, sleep showed significantly reduced functional connectivity for the whole brain compared to that during wakefulness. REM sleep showed statistically different connectivity patterns compared to non-REM sleep in sleep-related subcortical brain circuits. Conclusion: This study suggests the feasibility of exploring functional brain networks using sleep EEG-fMRI for whole night sleep via appropriate experimental procedures and signal processing techniques for fMRI and EEG.

• The KIPS Transactions:PartA
• /
• v.17A no.3
• /
• pp.113-120
• /
• 2010

Even networks were introduced as a class of fault-tolerant multiprocessor networks and analyzed so many useful properties and algorithms such as simple routing algorithms, maximal fault tolerance, node disjoint path. Introduced routing algorithms and node disjoint path algorithms are proven to be optimal. However, it has not been introduced to constructing scheme for edge-disjoint spanning trees in even networks. The design of edge-disjoint spanning trees is a useful scheme to analyze for measuring the efficiency of fault tolerant of interconnection network and effective broadcasting. Introduced routing algorithm or node disjoint path algorithm are for the purpose of routing or node disjoint path hence they are not applicable to constitute edge disjoint spanning tree. In this paper, we show a construction algorithm of edge-disjoint spanning trees in even network $E_d$.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.2
• /
• pp.335-344
• /
• 2020

This study compared the structures and changes of help network between pre-service secondary teachers and pre-journalists who participated in the class related to network analysis to seek ways to activate a professional learning community. For this study, we used Exponential Random Graph Models (ERGM) based on ties as being interdependent and not conventional regression models requiring assumptions of independence between observations. The analysis subjects were 43 pre-service secondary teachers and 29 pre-journalists who responded both early and late in the help network survey. The main results were as follows. First, full models with network structural terms were better than simple models with no structural terms. Second, the effect of transitivity was not statistically significant in the pre-service secondary teachers' network. However, it was statistically significant in the pre-journalists' network. Third, there were effects of reciprocity, indegree popularity, and outdegree activity in the early help network of pre-service secondary teachers. On the contrary, there were only the positive effects of reciprocity and the negative effect of outdegree activity in the late network. Finally, this study demonstrated the possibility in educational fields' application of network structural effects and provided limitations and directions for future research.

• Journal of Korea Multimedia Society
• /
• v.12 no.1
• /
• pp.139-146
• /
• 2009

This paper deals with visualization that can appropriately show the characteristics deduced from relationships between pieces of information. The visualization of influence, which is used as an important index in deducing the characteristics of relationships in social network analysis, was selected as research topic, and first, the elements that show relationships within the network and the index that show influence were classified and organized. Second, the links between relational elements that show influence in social network were examined, and an influence visualization network was created. Third, an influence visualization framework was proposed which explains the interaction between social network analysis and visualization process. The influence visualization network and framework being proposed in this paper can be used not only to understand and analyze the elements that influence social network but also to make it possible to have a rational and efficient approach to network visualization. Hopefully, they will become a new methodological approach to information design.

• The KIPS Transactions:PartA
• /
• v.11A no.1
• /
• pp.49-56
• /
• 2004

The performance of the multiprocessor systems is limited by the several factors. The system performance is affected by the processor speed, memory delay, and interconnection network bandwidth/latency. By the evolution of semiconductor technology, off the shelf microprocessor speed breaks beyond GHz, and the processors can be scalable up to multiprocessor system by connecting through the interconnection networks. In this situation, the system performances are bound by the latencies and the bandwidth of the interconnection networks. SCI, Myrinet, and Gigabit Ethernet are widely adopted as a high-speed interconnection network links for the high performance cluster systems. Performance improvement of the interconnection network can be achieved by the bandwidth extension and the latency minimization. Speed up of the operation clock speed is a simple way to accomplish the bandwidth and latency betterment, while its physical distance makes the difficulties to attain the high frequency clock. Hence the system performance and scalability suffered from the interconnection network limitation. Duplicating the link of the interconnection network is one of the solutions to resolve the bottleneck of the scalable systems. Dual-ring SCI link structure is an example of the interconnection network improvement. In this paper, I propose a network topology and a transaction path algorism, which optimize the latency and the efficiency under the duplicated links. By the simulation results, the proposed structure shows 1.05 to 1.11 times better latency, and exhibits 1.42 to 2.1 times faster execution compared to the dual ring systems.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1995.06a
• /
• pp.73-75
• /
• 1995

연속 음성 인식을 위한 신경망과 학습속도를 줄이기 위한 연결강도 초기화에 관해 다루고 있다. 우선 음소를 여러개의 그룹으로 나눈 후 각각의그룹에 대한 음소를 인식하는 신경망과 자신의 그룹을 판별하는 VGNN 으로 신경망을 구성한다. 여기서 구성되는 신경망은 각각의 음소를 인식하는 출력을 낼 뿐 아니라, 입력이 자신의 그룹에 속하는지 그렇지 않은지를 판별하는 출력을 낸다. 이런 신경망을 학습시키는 데 상당한 시간이 걸리므로 이 신경망의 학습속도를 줄이기 위해 학습 데이터를 사용하여 신경망의 연결 강도를 초기화한다.