• 한국음향학회지
• /
• 제14권1호
• /
• pp.13-21
• /
• 1995

본 논문에서는 신경망 학습에 의한 영상분류 알고리즘을 개선하였으며, 이것은 입력패턴 생성부와 분류을 위한 역전파 알고리즘에 의한 광역신경망으로 구성된다. 입력패턴을 위한 특징값으로는 자기조직화 형상지도 학습에 의해 얻은 코드북 데이타를 특징벡터로 이용한다. 이것은 입력벡터로서 원영상에 충실하면서 입력 뉴런수를 감소시킨다. 분류기에 사용된 광역망 알고리즘은 가중치와 유니트 오프셋 제어가 가능하도록 역전파 알고리즘에 제어부와 어드레스 메모리부를 삽입하였다. 실험결과 이들 분류기는 학습시 국소최소점에 빠지지 않게 되며, 대규모 신경망을 구현하고자 할 때 망구조를 간단히 할 수 있다. 또한 이것은 동작속도를 크게 개선할 수 있다.

• 전자공학회논문지
• /
• 제51권9호
• /
• pp.165-170
• /
• 2014

음성 인식을 위한 전처리기로 주변 잡음을 제거해 주는 음성향상 기법이 강조되고 있다. 다양한 음성향상 기법들 중 코드북 기반 음성향상 기법은 nonstationary 잡음 환경에서도 효율적으로 동작한다. 하지만, 기존 코드북 기반 음성향상 기법에서는 입력 신호와 음성 및 잡음 코드벡터 간에 미스매치가 발생하여 부정확한 게인이 추정되는 문제가 있다. 본 논문에서는 부정확한 게인을 보상하기 위해 long-term 잡음 추정 알고리즘을 사용하여 매 프레임 별로 신호 대 잡음비기반의 Normalized Weighting Factor (NWF)를 구하고, 이것을 기존 게인에 보상하는 방식을 제안한다. 제안된 코드북 기반 음성향상 기법은 기존 코드북 기반 음성향상 기법에 비해 향상된 성능을 보였다.

• Bulletin of the Korean Chemical Society
• /
• 제31권8호
• /
• pp.2215-2218
• /
• 2010

The series of alkali metal perhalogenates, $MXO_4$ (M=Li, Na, K and X=F, Cl, Br, I) were theoretically studied with the help of MP2 methods. Bidentate as well as tridentate structures were found to be stable minima. The bidentate structures are becoming preferred as the size of halogen increases and as the size of metal decreases. Geometrically, the M-O and M-X distances of both bidentate and tridentate structures, increase with the size of metal. Generally, the M-$O_1$ distances of tridentate forms are longer than the corresponding distances of bidentate forms, while the M-X distances of tridentate forms show the opposite trend. Similarly, the X-O bonds increase with the size of halogens except $MXO_4$ pairs, where the X-O bonds are unusually long due to the enhanced oxygen-oxygen repulsions. In short, the relative energetics as well as the geometrical parameters are found to be strongly dependent on halogen and metal elements.

• Bulletin of the Korean Chemical Society
• /
• 제28권12호
• /
• pp.2343-2353
• /
• 2007

We report the application of time-dependent density functional theory (TDDFT) to the calculation of potential energy profile relevant to the excited state intramolecular proton transfer (ESIPT) processes in title molecules. The TDDFT single point energy calculations along the reaction path have been performed using the CIS optimized structure in the excited state. In addition to the Stokes shifts, the transition energies including absorption, fluorescence, and 0-0 transition are estimated from the TDDFT potential energy profiles along the proton transfer coordinate. The excited state TDDFT potential energy profile of SA and 3ASA resulted in very flat function of the OH distance in the range ROH = 1.0-1.6 A, in contrast to the relatively deep single minimum function in the ground state. Furthermore, we obtained very shallow double minima in the excited state potential energy profile of SA and 3ASA in contrast to the single minimum observed in the previous work. The change of potential energy profile along the reaction path induced by the substitution of electron donating groups (-NH2 and -OCH3) at different sites has been investigated. Substitution at para position with respect to the phenolic OH group showed strong suppression of excited state proton dislocation compared with unsubstitued SA, while substitution at ortho position hardly affected the shape of the ESIPT curve. The TDDFT results are discussed in comparison with those of CASPT2 method.

• 토지주택연구
• /
• 제9권4호
• /
• pp.1-9
• /
• 2018

Since year 2000, French housing and rent prices rose at a rapid rate and the housing market has been overheated. Face to this phenomena, the French government enacted a new law Alur which is a legislatif tool to control the private housing rent price for the cities, where the tension of the housing market is very high. This new law has impacted the housing market in two major ways. First, for the 38 cities designated by this law, the rent price's increase rate can not rise above the IRL, which is the rent reference index. Secondly, this law also permits local authorities to control the housing rent's price following the concrete price guidance. Especially in Paris, the city applicated this method for private rental housing since 2015. This city classified its own area by 14 zones. Based on the market surveys of each sector, local authority made a guidance for private housing rent's price. The guideline is consisted of average prices, maxima and minima price by types, which is classified by the construction year, number of rooms and furnished or not. Therefore, this study aims to understand french housing rent's price control system and draw implementation for korean housing rent policies. This research is meaningful for it introduces recent foreign regislations which could be helpful to control the housing market in Korea.

• 천문학회보
• /
• 제41권2호
• /
• pp.62.2-62.2
• /
• 2016

We present a newly developed algorithm based on a Bayesian method for 2D tilted-ring analysis of disk galaxies which operates on velocity fields. Compared to the conventional ones based on a chi-squared minimisation procedure, this new Bayesian-based algorithm less suffers from local minima of the model parameters even with high multi-modality of their posterior distributions. Moreover, the Bayesian analysis implemented via Markov Chain Monte Carlo (MCMC) sampling only requires broad ranges of posterior distributions of the parameters, which makes the fitting procedure fully automated. This feature is essential for performing kinematic analysis of an unprecedented number of resolved galaxies from the upcoming Square Kilometre Array (SKA) pathfinders' galaxy surveys. A standalone code, the so-called '2D Bayesian Automated Tilted-ring fitter' (2DBAT) that implements the Bayesian fits of 2D tilted-ring models is developed for deriving rotation curves of galaxies that are at least marginally resolved (> 3 beams across the semi-major axis) and moderately inclined (20 < i < 70 degree). The main layout of 2DBAT and its performance test are discussed using sample galaxies from Australia Telescope Compact Array (ATCA) observations as well as artificial data cubes built based on representative rotation curves of intermediate-mass and massive spiral galaxies.

• 한국지능시스템학회논문지
• /
• 제5권3호
• /
• pp.83-91
• /
• 1995

본 연구에서는 종래에 학습법으로 널리 이용되고 있는 역전파학습법의 문제점으로 지적되어 온 학습에 많은 시간이 걸리는 점과 국소적 최적해에 해가 수렴하여 오차가 충분히 작게 되지 않는 등의 문제점을 해결하기 위해, Hu에 의해 고안된 random tabu 탐색법을 이용하여 신경회로망의 연결강도를 최적화하는 학습알고리즘을 새로이 제안하였다. 그리고 이 방법을 배타적 논리합 문제에 적용하여 기존의 역전파학습법과 학습상수 $, $에 tabu탐색법을 이용한 결과와 비교 검토하여 본 방법이 국소적 최적해에 수렴하지 않고 수렴정도를 개선할 수 있음을 확인하였다.

• ETRI Journal
• /
• 제22권1호
• /
• pp.38-46
• /
• 2000

The extraction of model parameters for embedded passive components is crucial for designing and characterizing the performance of multichip module (MCM) substrates. In this paper, a method for optimizing the extraction of these parameters using genetic algorithms is presented. The results of this method are compared with optimization using the Levenberg-Marquardt (LM) algorithm used in the HSPICE circuit modeling tool. A set of integrated resistor structures are fabricated, and their scattering parameters are measured for a range of frequencies from 45 MHz to 5 GHz. Optimal equivalent circuit models for these structures are derived from the s-parameter measurements using each algorithm. Predicted s-parameters for the optimized equivalent circuit are then obtained from HSPICE. The difference between the measured and predicted s-parameters in the frequency range of interest is used as a measure of the accuracy of the two optimization algorithms. It is determined that the LM method is extremely dependent upon the initial starting point of the parameter search and is thus prone to become trapped in local minima. This drawback is alleviated and the accuracy of the parameter values obtained is improved using genetic algorithms.

• 천문학회지
• /
• 제44권3호
• /
• pp.97-108
• /
• 2011

The photometric light curves of the W-type W UMa eclipsing contact binary system BB Pegasi have been found to be extremely asymmetric over all the observed 63 years in all wavelengths UBVR. The light curves have been characterized by occultation primary minima. Hence, the morphology of these light curves has been studied in view of these different asymmetric degrees. The system shows a distinct O'Connell effect, as well as depth variation. A 22.96 years of stellar dark spots cycle has been determined for the system. Almost the same cycle (22.78 yr) has been found for the depth variation of MinI and MinII. We also present an analysis of mid-eclipse time measurements of BB Peg. The analysis indicates a period decrement of $5.62{\times}10^{-8}$ day/yr, which can be interpreted in terms of mass transfer at a rate of $-4.38{\times}10^{-8}M_{\odot}$/yr, from the more to the less massive component. The O - C diagram shows a damping sine wave covering two different cycles of 17.0 yr and 12.87 yr with amplitudes equal to 0.0071 and 0.0013 day, respectively. These unequal durations show a non-periodicity which may be explained as a result of magnetic activity cycling variations due to star spots. The obtained characteristics are consistent with similar chromospherically active stars, when applying the Applegate's (1992) mechanism.

• 한국생물정보학회:학술대회논문집
• /
• 한국생물정보시스템생물학회 2004년도 The 3rd Annual Conference for The Korean Society for Bioinformatics Association of Asian Societies for Bioinformatics 2004 Symposium
• /
• pp.250-261
• /
• 2004

A novel method for ab initio prediction of protein tertiary structures, PROFESY (PROFile Enumerating SYstem), is introduced. This method utilizes secondary structure prediction information and fragment assembly. The secondary structure prediction of proteins is performed with the PREDICT method which uses PSI-BLAST to generate profiles and a distance measure in the pattern space. In order to predict the tertiary structure of a protein sequence, we assemble fragments in the fragment library constructed as a byproduct of PREDICT. The tertiary structure is obtained by minimizing the potential energy using the conformational space annealing method which enables one to sample diverse low lying minima of the energy function. We apply PROFESY for prediction of some proteins with known structures, which shows good performances. We also participated in CASP5 and applied PROFESY to new fold targets for blind predictions. The results were quite promising, despite the fact that PROFESY was in its early stage of development. In particular, the PROFESY result is the best for the hardest target T0161.