• Journal of Sensor Science and Technology
• /
• v.8 no.1
• /
• pp.45-52
• /
• 1999

In this raper, an enhanced learning method is proposed for improving the learning speed of the error back propagation learning algorithm. In order to cope with the premature saturation phenomenon at the initial learning stage, a variation scheme of active functions is introduced by using higher order functions, which does not need much increase of computation load. It naturally changes the learning rate of inter-connection weights to a large value as the derivative of sigmoid function abnormally decrease to a small value during the learning epoch. Also, we suggest the hybrid learning method incorporated the proposed method with the momentum training algorithm. Computer simulation results show that the proposed learning algorithm outperforms the conventional methods such as momentum and delta-bar-delta algorithms.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.649-651
• /
• 2018

본 논문에서는 CNN에서 클래스 활성화 맵과 원샷 러닝을 결합하여 트위터 분류를 위한 딥 러닝 모델을 제안한다. 클래스 활성화 맵은 트윗 분류에 대한 분류 주제와 연관된 핵심 어휘를 추출하고 강조 표시하도록 사용되었다. 특히 작은 학습 데이터 셋을 사용하여 다중 클래스 분류의 성능을 향상시키기 위해 원샷 러닝 방법을 적용한다. 제안하는 방법을 검증하기위해 TREC 2018 태스크의 사건 스트림(TREC-IS) 학습데이터를 사용하여 비교실험을 했다. 실험 결과에서 CNN 기본 모델의 정확도는 58.1%이고 제안 방법의 정확도는 69.6%로 성능이 향상됨을 보였다.

• Proceeding of EDISON Challenge
• /
• 2017.03a
• /
• pp.719-725
• /
• 2017

막전압 고정 기법은 세포막 이온통로의 활성화 게이트, 비활성화 게이트의 물리적 성질 등을 밝힐 수 있다. 즉, 여러 다양한 펄스 프로토콜을 이용하여 활성화 게이트와 비활성화 게이트의 막전압 의존성을 구할 수 있다. 본 연구는 L-type $Ca^{2+}$ 통로의 모델을 막전압 고정 기법 시뮬레이션에 적용하여 최적의 펄스 프로토콜을 얻기 위한 방법을 제시하고자 하였다. 비활성화 게이트의 막전압 의존성을 구하는 경우, 테스트 전압에서 +10 mV의 전압으로 가기 전에 0 ms, 5 ms, 20 ms의 gap을 주었는데 이 중 5 ms의 gap을 주었을 때 모델과 가장 가까운 관계를 얻을 수 있었다. 다음으로 활성화 게이트의 막전압 의존성을 구하는 경우, 일반적인 방법으로는 실제 관계와 크게 다른 결과를 얻었으나, 0 mV 이하의 막전압에 대해서만 막전압 의존성을 구하는 방법을 사용하여 실제 관계와 근접한 결과를 얻을 수 있었다. 따라서, 본 시뮬레이션 프로그램을 적절히 이용한다면 실제 세포실험에서 정확한 수치를 얻기 위한 펄스 프로토콜을 얻는데 활용할 수 있을 것으로 본다.

• 한국신재생에너지학회:학술대회논문집
• /
• 2007.11a
• /
• pp.673-676
• /
• 2007

교토의정서의 발효로 인해 전세계적으로 온실가스 감축 활동이 활발하게 진행되고 있으며, 국내 산업계 또한 온실가스 저감을 피할 수 없게 되었다. 특히 교토의정서 이행을 위해 도입된 교토메카니즘 가운데 개발도상국의 온실가스 감축 활동을 위해 CDM 사업이 활성화되고 있으며, 개발도상국인 우리나라도 참여할 수 있기 때문에 국내 기업 및 정부의 관심이 높아지고 있다. 그러나 CDM 사업을 수행하기 위해서는 적용 가능한 CDM 사업 방법론이 필요하며, 방법론이 없을 경우 CDM 사업 방법론을 개발하여야 한다. 방법론에는 베이스라인, 추가성, 배출 감축량 등에 대한 구체적인 방법이 제시되어야 하고 사업의 타당성 확인 및 검증을 위해 방법론을 정확히 이해해야 한다. 따라서 본 CDM 사업 방법론 연구를 통해 CDM 사업 수행 및 방법론 개발을 위해 필요한 주요 항목에 대한 이해를 돕고 국내 CDM 사업의 활성화를 도모하고자 한다.

• Journal of Korean Society of Transportation
• /
• v.28 no.2
• /
• pp.87-98
• /
• 2010

Recently active policies of bicycle in many municipalities have been selected as the main policies. And a bike-related research is a difference in the direction of research according to socially active atmosphere on a bike using. Recent the research of approaches to enable bike use according to regional characteristics focused on Busan has been conducted is differentiated. But A different approach should be considered on the small town smaller than Metropolitan. Because of the relatively small city-size bicycle can be a variety of purposes. Therefore, this study to research awareness assessment methods of these local residents on the bike use inquires to the awareness of local residents in addition to the physical condition of facilities and policies target for the dong unit of small towns with 300,000~500,000 people. Awareness assessment methods of these local residents for the revitalization of bicycle using enable the local and regional conditions of bicycle using to compare very easily and the conception on the revitalization of bicycles using will make it possible considering the regional differences in the conditions of dong unit. Therefore, this process will be important references to be considered a differentiated analysis and approaches to enable bike use of the conditions of small towns on the bicycle using.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.3A
• /
• pp.304-310
• /
• 2006

This paper evaluates the performance of three types of data delivery mechanisms in clustered sensor networks, as a basic research to develop an energy efficient topology management scheme. In the first mechanism, one node per cluster(clusterhead) turns on its radio(or wakes up) to transmit and receive RTS/CTS/DATA/ACK messages, but in the second one, k nodes per cluster wake up and participate in the message exchange. In the last mechanism, clusterheads turn on the radio to exchange RTS/CTS messages, and if a clusterhead receives RTS containing its cluster m as a destination, it makes k nodes in the cluster hun on the radio to receive DATA and transmit ACK. Through simulation, we show the energy consumption of the three types of data delivery mechanisms as functions of the number of active nodes per cluster, offered load, and packet loss probability.

• Journal of the Korean Electrochemical Society
• /
• v.12 no.1
• /
• pp.40-46
• /
• 2009

We present a method for spatially selective immobilization of functional materials, such as proteins and nanoparticles, onto pre-activated silicon surfaces by electrochemical reaction. Carboxymethylbenzendiazonium (CMBD) cations, being adsorbable on silicon surfaces through electrochemically reductive deposition, is used as an anchor molecule to prepare the pre-activated silicon surfaces. It is demonstrated that the use of BD reaction is very efficient for the selective immobilization because the functional materials are immobilized exclusively onto the pre-adsorbed CMBD region. The method is applied to immobilize gold nanoparticles on the selected nanowire of the nanowire array.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.5
• /
• pp.187-193
• /
• 2020

The purpose of this study is to study the methodology of deriving a policy that activates artificial intelligence from the governmental perspective in order to induce corporate growth by effectively grafting artificial intelligence technology into society and thereby improve individual and national competitiveness by creating new jobs. In order to derive activation plans, 1) detailed investigation of the domestic environment, 2) discovery of priority support fields and models that can be applied to artificial intelligence, 3) preparation of guidelines for activation and introduction, 4) specific methods for promoting and activating artificial intelligence Should be presented. The proposed artificial intelligence activation method performs a procedure to verify and confirm the effectiveness of artificial intelligence nurturing through a multi-faceted approach. The multi-faceted analysis approach includes business ecosystem aspects, industry-specific aspects including companies, technology fields, policy aspects, public and non-public services aspects, government-led and private-led aspects. Therefore, it can be reviewed as a method of inducing activation in various forms. In the future research field, it is necessary to prove the effectiveness of the proposed activation plan based on empirical data on artificial intelligence-based services. The expected effect of this study is to contribute to support the development of artificial intelligence technology and to establish related policies.

• Journal of the Korean Institute of Telematics and Electronics D
• /
• v.35D no.10
• /
• pp.83-90
• /
• 1998

The effects of dopant activation anneal on GOI (Gate Oxide Integrity) of MOS capacitor with amorphous silicon gate electrode were investigated. It was found that the amorphous silicon gate electrode was crystallized and the dopant atoms were sufficiently activated by activation anneal. The mechanical stress of gate electrode that reveals large compressive stress in amorphous state, was released with increase of anneal temperature from $700^{\circ}C$ to 90$0^{\circ}C$. The resistivity of gate electrode polycrystalline silicon film is decreased by the increase of anneal temperature. The reliability of thin gate oxide and interface properties between oxide and silicon substrate greatly depends on the activation anneal temperature. The charge trapping characteristics as well as oxide reliability are improved by the anneal of 90$0^{\circ}C$ compare to that of $700^{\circ}C$ or 80$0^{\circ}C$. Especially, the lifetimes of the thin gate oxide estimated by TDDB method is 3$\times$10$^{10}$ for the case of $700^{\circ}C$ anneal, is significantly increased to 2$\times$10$^{12}$ for the case of 90$0^{\circ}C$ anneal. Finally, the interface trap density is reduced with relaxation of mechanical stress of gate electrode.

• Journal of the Korean Vacuum Society
• /
• v.17 no.6
• /
• pp.544-548
• /
• 2008

At the semiconductor industry, Photoresist(PR) strip progress has high cost and time consuming process. Accordingly, many research group have been focused on the shortening of the PR strip progress. But the replacements of newly developed materials rather than normally used strip have accompanied by cost consumption. Therefore, we suggested the Plasma Liquid-Vapor Activation (PLVA) method of general PR strip solution for saving the PR strip time and the high strip rate of PR residue. The PLVA method was very effective for PR strip progress. Also, the ion damaged PR(high dose implanted photoresist: HDI-PR) was almost impossible to strip. However, it was very difficult to characterize the change of chemical composition of HDI-PR between with and without PLVA method. Thus, physical properties of HDI-PR surface with and without PLVA method were measured by using the nano-indenter system.