• Title/Summary/Keyword: 트리성장

Search Result 112, Processing Time 0.028 seconds

Regression Neural Networks for Improving the Learning Performance of Single Feature Split Regression Trees (단일특징 분할 회귀트리의 학습성능 개선을 위한 회귀신경망)

  • Lim, Sook;Kim, Sung-Chun
    • Journal of the Korean Institute of Telematics and Electronics B
    • /
    • v.33B no.1
    • /
    • pp.187-194
    • /
    • 1996
  • In this paper, we propose regression neural networks based on regression trees. We map regression trees into three layered feedforward networks. We put multi feature split functions in the first layer so that the networks have a better chance to get optimal partitions of input space. We suggest two supervised learning algorithms for the network training and test both in single feature split and multifeature split functions. In experiments, the proposed regression neural networks is proved to have the better learning performance than those of the single feature split regression trees and the single feature split regression networks. Furthermore, we shows that the proposed learning schemes have an effect to prune an over-grown tree without degrading the learning performance.

  • PDF

A study on the effect of growth of plant for the sound-barrier development (생태방음벽개발을 위한 식물종의 성장에 관한 연구)

  • Cho, Hae-Yong;Choi, Chang-Ha;Bashyal, Sarita;Lee, Byoung-Kyu;Han, Se-Gwon;Shim, Wang-Sub
    • Proceedings of the KAIS Fall Conference
    • /
    • 2008.11a
    • /
    • pp.375-378
    • /
    • 2008
  • 생태방음벽의 적합한 식물종은 최소한의 유지관리만을 통해 성장할 수 있는 식물을 선정하는 것이 매우 중요하다. 지상부의 성장이 억제되고, 지하부의 성장이 활발해야 하기 때문에 레볼센트리, 소나타, 페난트 등이 가장 적합한 식물종으로 판단된다. 코스모스의 경우는 발아율, 생명력, 번식력 등이 매우 좋으나 지하부의 성장이 다른 품종에 비해 약 35%이상 크게 성장하고 지하부의 성장이 50%미만이기 때문에 식생방음벽으로서는 적합하지 않다. 그리고 여러 조건을 충족할 수 있는 추후 실험이 이루어져야 할 것으로 사료된다.

  • PDF

Detection of Ultrasonic Signals Associated with Tree Growth in Epoxy resin (애폭시수지의 트리성장에 따른 초음파 신호검출)

  • 이상우;송현직;이광식;이동인;김인식
    • Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
    • /
    • 1998.06a
    • /
    • pp.285-288
    • /
    • 1998
  • Ultrasonic signal characteristics related with the growth of electrical trees in epoxy resin are examined under 67[Hz] ac voltage application along with a CCD camera and PD current method. The ultrasonic sensor with a resonant frequency of 200[Hz] supported by a mechanical spring is attached directly to the lower-side of plane electrode. The magnitude of Partial discharge and count rate of ultrasonic signals have been measured according to the tree growth from a needle tip in an epoxy sample.

  • PDF

AC Breakdown Voltage Characteristics of 22.9kV TR XLPE Power Cable Before and After 14days Cyclic Aging and AWTT (14주기 열화 및 AWTT에 따른 22.9kV TR CNCV-W 전력케이블의 교류파괴전압 특성분석)

  • Kim, We-Young;Heo, Jong-Cheol;Park, Tae-Gone
    • Proceedings of the KIEE Conference
    • /
    • 2006.07c
    • /
    • pp.1507-1508
    • /
    • 2006
  • XLPE 절연 전력케이블의 제조과정에서 발생하는 가교부산물을 제거할 목적으로 14주기노화를 실시하며, XLPE 절연체에 발생하는 수트리를 가속 발생 성장시킬 목적으로 가속수트리 열화를 실시한다. 22.9kV 트리억제형(TR CNCV-W) 케이블에 대하여 14주기노화 전과 후, 120, 240, 360일 가속수트리(AWTT) 후의 파괴전압을 분석하였다. 14주기노화는 상승효과와 감소효과가 비슷하게 작용하는 것으로 나타났으며, 240일 AWTT 이후에 노화가 가속되는 것으로 나타났다.

  • PDF

Frequently Occurred Information Extraction from a Collection of Labeled Trees (라벨 트리 데이터의 빈번하게 발생하는 정보 추출)

  • Paik, Ju-Ryon;Nam, Jung-Hyun;Ahn, Sung-Joon;Kim, Ung-Mo
    • Journal of Internet Computing and Services
    • /
    • v.10 no.5
    • /
    • pp.65-78
    • /
    • 2009
  • The most commonly adopted approach to find valuable information from tree data is to extract frequently occurring subtree patterns from them. Because mining frequent tree patterns has a wide range of applications such as xml mining, web usage mining, bioinformatics, and network multicast routing, many algorithms have been recently proposed to find the patterns. However, existing tree mining algorithms suffer from several serious pitfalls in finding frequent tree patterns from massive tree datasets. Some of the major problems are due to (1) modeling data as hierarchical tree structure, (2) the computationally high cost of the candidate maintenance, (3) the repetitious input dataset scans, and (4) the high memory dependency. These problems stem from that most of these algorithms are based on the well-known apriori algorithm and have used anti-monotone property for candidate generation and frequency counting in their algorithms. To solve the problems, we base a pattern-growth approach rather than the apriori approach, and choose to extract maximal frequent subtree patterns instead of frequent subtree patterns. The proposed method not only gets rid of the process for infrequent subtrees pruning, but also totally eliminates the problem of generating candidate subtrees. Hence, it significantly improves the whole mining process.

  • PDF

Epitaxial growth and microstructural characterization of $YSi_2$ films on (100)Si substrate (이트리움 실리사이드 박막의 (100)Si 기판상에서의 방향성 성장과 미세조직의 특성)

  • Lee, Young-Ki
    • Journal of the Korean Crystal Growth and Crystal Technology
    • /
    • v.7 no.1
    • /
    • pp.59-69
    • /
    • 1997
  • The growth and microstructural characterization of epitaxial yttrium silicide ($YSi_2$) on the (100)Si substrate are investigated. The $YSi_2$ film grow epitaxilly through the solid phase reaction during vacuum annealing above $400^{\circ}C$. The epitaxial relationships between the hexagonal $YSi_2$ film and the (100)Si substrate are [0001]$YSi_2$//[011]Si and [0001]$YSi_2$//[011]Si in the (1100)$YSi_2$//(100)Si plane relation. The YSi$_2$ film consists of the two types of domains which have two different azimuthal orientations making an angle of $90^{\circ}$ to each other. The two types of domains in the $YSi_2$film are equivalent in volume fraction and crystalline quality, which has been proved from the equivalent integrated intensities of (2201) asymmetric reflection of X-ray diffraction. The formation of a double -domain structure is discussed on the basis of geometrical matching at interface between the (1100)$YSi_2$ film and the (100)Si substrate, and growth model is proposed.

  • PDF

Development of Reuse Process Through Recovery and Refinement of Precursor for LED (LED용 precursor 재이용을 위한 회수 및 정제 공정 개발)

  • Yang, Jae Yeol;O, Byung Sung;Yoon, Jae Sik
    • Resources Recycling
    • /
    • v.23 no.1
    • /
    • pp.25-32
    • /
    • 2014
  • The purpose of this research is to develop a process and a system to collect, purify and reuse the residual quantity of trimethylgallium, used as a raw material, upon GaN epitaxial growth for LED from a metal organic chemical vapor deposition(MOCVD) equipment. This research reviews whether TMGa collected from the process can be used through a chemical and structural characteristics evaluation. As a result of analyzing the purity using ICP-MS and ICP-AES, 7N high purity (99.99999%) of TMGa was obtained. According to checking the structural change of TMGa through NMR analysis, TMGa having pure $(CH_3)_3Ga$ structure was obtained without structural change. For reliability review of the collected TMGa, u-GaN was deposited using the MOCVD process and an structural, optical and electrical characteristics evaluation was conducted. As a result, it was found out that the reuse was possible.

An Improved Function Synthesis Algorithm Using Genetic Programming (유전적 프로그램을 이용한 함수 합성 알고리즘의 개선)

  • Jung, Nam-Chae
    • Journal of the Institute of Convergence Signal Processing
    • /
    • v.11 no.1
    • /
    • pp.80-87
    • /
    • 2010
  • The method of function synthesis is essential when we control the systems not known their characteristic, by predicting the function to satisfy a relation between input and output from the given pairs of input-output data. In general the most systems operate non-linearly, it is easy to come about problem is composed with combinations of parameter, constant, condition, and so on. Genetic programming is proposed by one of function synthesis methods. This is a search method of function tree to satisfy a relation between input and output, with appling genetic operation to function tree to convert function into tree structure. In this paper, we indicate problems of a function synthesis method by an existing genetic programming propose four type of new improved method. In other words, there are control of function tree growth, selection of local search method for early convergence, effective elimination of redundancy in function tree, and utilization of problem characteristic of object, for preventing function from complicating when the function tree is searched. In case of this improved method, we confirmed to obtain superior structure to function synthesis method by an existing genetic programming in a short period of time by means of computer simulation for the two-spirals problem.

Design Of Compact Passive Triplexer Module for Wi-MAX Application (수동 소자를 이용한 Wi-MAX용 트리플렉서 모듈 설계)

  • Jeon, Jae-W.;Seo, Jae-H.;Lee, Seung-C.;Lee, Sang-A.;Lim, Jong-H.;Cheon, Seong-J.;Park, Jae-Y.
    • Proceedings of the KIEE Conference
    • /
    • 2009.07a
    • /
    • pp.2253_2254
    • /
    • 2009
  • 무선 모바일 통신 시장의 성장에 따라 수동 소자로 구성된 모듈에 대한 관심이 높아지고 있다. 본 논문에서는 수동소자를 이용한 Wi-MAX용 트리플렉서를 설계하였다. 트리플렉서는 Wi-MAX의 RF Front-end단 앞부분에 연결되며, IEEE 802.16에 따른 2GHz, 3GHz, 5GHz 대역을 선택, 분리해주는 역할을 한다. 제안된 트리플렉서는 저역통과필터, 대역통과필터, 고역통과필터로 구성하였으며, 수동소자의 최소화로 인해 삽입손실 향상을 중점으로 설계하였다. 회로해석 결과, 각 대역별 (2/3/5GHz) 삽입손실은 각각 -0.5dB, -0.6dB, -0.4dB 였으며, 반사손실은 -20dB, -20dB, -12dB, 격리도는 -20dB, -24dB, -25dB의 특성을 보였다. 수동 소자의 최소화 설계로 인해 모듈 및 시스템의 저가화를 기대할 수 있다.

  • PDF

A Two-Dimensional Binary Prefix Tree for Packet Classification (패킷 분류를 위한 이차원 이진 프리픽스 트리)

  • Jung, Yeo-Jin;Kim, Hye-Ran;Lim, Hye-Sook
    • Journal of KIISE:Information Networking
    • /
    • v.32 no.4
    • /
    • pp.543-550
    • /
    • 2005
  • Demand for better services in the Internet has been increasing due to the rapid growth of the Internet, and hence next generation routers are required to perform intelligent packet classification. For a given classifier defining packet attributes or contents, packet classification is the process of identifying the highest priority rule to which a packet conforms. A notable characteristic of real classifiers is that a packet matches only a small number of distinct source-destination prefix pairs. Therefore, a lot of schemes have been proposed to filter rules based on source and destination prefix pairs. However, most of the schemes are based on sequential one-dimensional searches using trio which requires huge memory. In this paper, we proposea memory-efficient two-dimensional search scheme using source and destination prefix pairs. By constructing binary prefix tree, source prefix search and destination prefix search are simultaneously performed in a binary tree. Moreover, the proposed two-dimensional binary prefix tree does not include any empty internal nodes, and hence memory waste of previous trio-based structures is completely eliminated.