• Journal of information and communication convergence engineering
• /
• 제8권3호
• /
• pp.267-272
• /
• 2010

For the last decade, recurrent neural networks (RNNs) have been commonly applied to communications channel equalization. The major problems of gradient-based learning techniques, employed to train recurrent neural networks are slow convergence rates and long training sequences. In high-speed communications system, short training symbols and fast convergence speed are essentially required. In this paper, the derivative-free Kalman filter, so called the unscented Kalman filter (UKF), for training a fully connected RNN is presented in a state-space formulation of the system. The main features of the proposed recurrent neural equalizer are fast convergence speed and good performance using relatively short training symbols without the derivative computation. Through experiments of nonlinear channel equalization, the performance of the RNN with a derivative-free Kalman filter is evaluated.

• 한국지능시스템학회논문지
• /
• 제15권5호
• /
• pp.552-559
• /
• 2005

실시간 순환형 훈련 알고리즘(RTRL)과 같이 경사법에 의해 훈련되는 순환형 뉴럴 네트웍(RNN)은 수렴속도가 매우 느린 단점을 지니고 있다. 이 알고리즘은 또한 오차 역전달 처리과정에서 결코 쉽지 않은 미분 계산을 필요로 한다. 본 논문에서는 완전하게 결합된 RNN의 훈련을 위하여 소위 언센티드 칼만필터라고 불리우는 미분없는 칼만필터 훈련 알고리즘을 시스템의 상태공간 상에서 표현하였다. 미분없는 칼만필터 훈련 알고리즘은 순환형 뉴럴 네트웍 훈련시 미분 계산 없이 매우 빠른 수렴속도와 좋은 추정 성능을 보여준다. 비선형 채널 등화 실험을 통하여 미분 없는 칼만필터 훈련 알고리즘을 이용한 RNN의 성능이 향상되었음을 보였다.

• 한국광학회지
• /
• 제25권5호
• /
• pp.273-278
• /
• 2014

일련의 국부영역으로부터 이차미분값을 획득하여 전체 형상을 복원하는 측정법을 제안하였다. 측정시 기울기가 제거된 국부형상에 대해 곡률과 이차미분이 동일시 되는 점을 이용하여, 3개의 이차미분값으로부터 직교하는 2방향을 따라 3차원형상을 복원할 수 있는 알고리즘을 구현하였다. 임의로 발생시킨 Zernike다항식의 계수로 자유곡면형상을 생성시키고, 구현된 알고리즘을 적용함으로써 검증과정을 수행하였다. 적용한 결과 최대 0.8 mm Sag를 갖는 직경 200 mm영역의 자유곡면형상에 대해 RMS 19 nm 형상복원오차를 갖고 복원됨을 확인하였다. 측정오차에 대한 복원오차 민감도를 진단하기 위해 SNR(Signal-to-Noise Ratio) 16의 가우시언 랜덤 노이즈를 부여한 후, 복원되는 형상의 오차를 진단한 결과, 197 nm의 형상복원오차가 발생함을 확인하였다.

• Advances in materials Research
• /
• 제3권2호
• /
• pp.117-128
• /
• 2014

Processable conducting polyaniline derivative with free amine functional groups was successfully synthesized from the monomer o-phenylenediamine in aqueous hydrochloric acid medium using ammonium persulfate as an oxidative initiator. The synthesized poly(o-phenylenediamine) (PoPD) in critical condition was found to be completely soluble in common organic solvents like dimethyl sulfoxide, N,N-dimethyl formamide etc. From the intrinsic viscosity measurement, the optimum condition for the polymerization was established. The polymer was characterized by ultraviolet visible spectroscopy, Fourier transform infrared spectroscopy, proton magnetic resonance spectroscopy ($^1HNMR$) and thermogravimetric (TGA) analyses. The weight average molecular weights of the synthesized polymers were determined by the dynamic light scattering (DLS) method. From the spectroscopic analysis the structure was found to resemble that of polyaniline derivative with free amine functional groups attached to ortho/meta position in the phenyl ring. However, very little ladder unit was also present with in the polymer chain. The moderate thermal stability of the synthesized polymer could be found from the TGA analysis. The average DC conductivity of $2.8{\times}10^{-4}S/cm$ was observed for the synthesized polymer pellet after doping with hydrochloric acid.

• Nonlinear Functional Analysis and Applications
• /
• 제26권4호
• /
• pp.839-853
• /
• 2021

Let 𝕽ⁿ be an Euclidean space and g : 𝕽ⁿ → 𝕽ⁿ be a monotone and continuous mapping. Suppose the convex constrained nonlinear monotone equation problem x ∈ 𝕮 s.t g(x) = 0 has a solution. In this paper, we construct an inertial-type algorithm based on the three-term derivative-free projection method (TTMDY) for convex constrained monotone nonlinear equations. Under some standard assumptions, we establish its global convergence to a solution of the convex constrained nonlinear monotone equation. Furthermore, the proposed algorithm converges much faster than the existing non-inertial algorithm (TTMDY) for convex constrained monotone equations.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2003년도 추계학술대회논문집
• /
• pp.814-818
• /
• 2003

This paper deals with the application of numerical differentiation in free vibration analysis. In the free vibration analysis, the derivative values of the given function are certainly used in calculation of structural parameters. For deriving the derivative values, both the time and labor are needed when the structures consist of non-linear geometries such as arches or curved beams. From this viewpoint, the numerical differentiation scheme is applied into the free vibration analysis. The numerical results obtained from the numerical differentiations are agreed very well with those obtained from the exact derivatives by analytical method. It is expected that the numerical differentiations can be utilized practically in the free vibration analysis.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 2006년도 정기 학술대회 논문집
• /
• pp.365-372
• /
• 2006

A new meshfree formulation is developed for material discontinuity problems. A local interfacial jump function which is defined as hyperplane function is embedded in the meshless approximation and the approximation accurately models functions with jumps in the displacement and the derivative fields. Diffuse derivative technique copes with difficulty due to complexity of derivative computation of meshfree approximation. Collocation method with diffuse derivative accelerates computing speed for numerical solution. By solving inclusion and composite material problems, the robustness and effectiveness of the method are verified.

• 한국응용약물학회:학술대회논문집
• /
• 한국응용약물학회 1994년도 춘계학술대회 and 제3회 신약개발 연구발표회
• /
• pp.272-272
• /
• 1994

신약개발에 있어서 대사연구는 약효의 species difference, 독성기전, 그리고 in vitro와 in vivo의 약효차이를 이해하는데 매우 중요한 역할을 담당하며 그 연구 영역을 넓혀가고 있다. 본 연구에서는 항혈전제로 개발 가능성이 있는 phosphodiesterase inhibitor인 KR30289의 rat과 rabbit에서 대사 연구를 통하여 약리효과의 상이점을 밝히고져 하였다. Mass analysis를 통하여 urinary 대사체의 구조를 규명한 결과 KR30289를 경구 투여 (1mg/kg)시 hydroxy derivative(M1)와 O-dealkylated derivative(M2)로 변환되어 urine으로 배설되었고 parent compound는 검출되지 않았다 M1과 M2 모두 free form, glucuronide conjugate, 그리고 sulfate conjugate 형태로 검출되었으나 rat과 rabbit 두 종간의 M1과 M2의 비율에서는 커다란 차이점을 보여주었다 rat의 경우 M2가 94% (free form : 14%, glucuronide conjugate 58%. sulfate conjugate, 22%)이었으며 rabbit에서는 M1이 54%로 상대적으로 많이 생성되었다(free form : 11%, glucuronide conjugate : 22%, sulfate conjugate 21%). 이와같이 두종간의 상이한 metabolic profile로 인하여 약리효과의 차이가 유발될수 있다고 추론된다.

• Advances in materials Research
• /
• 제12권2호
• /
• pp.101-118
• /
• 2023

The purpose of this paper is to study the effects of magnetic field and gravitational field on fiber-reinforced thermoelastic medium with memory-dependent derivative. Three-phase-lag model of thermoelasticity (3PHL) is used to study the plane waves in a fiber-reinforced magneto-thermoelastic material with memory-dependent derivative. A gravitating magneto-thermoelastic two-dimensional substrate is influenced by both thermal shock and mechanical loads at the free surface. Analytical expressions of the considered variables are obtained by using Laplace-Fourier transforms technique with the eigenvalue approach technique. A numerical example is considered to illustrate graphically the effects of the magnetic field, gravitational field and two types of mechanical loads(continuous load and impact load).

• International Journal of Fluid Machinery and Systems
• /
• 제8권3호
• /
• pp.209-219
• /
• 2015

A robust multi-fidelity optimization methodology has been developed, focusing on efficiently handling industrial runner design of hydraulic Francis turbines. The computational task is split between low- and high-fidelity phases in order to properly balance the CFD cost and required accuracy in different design stages. In the low-fidelity phase, a physics-based surrogate optimization loop manages a large number of iterative optimization evaluations. Two derivative-free optimization methods use an inviscid flow solver as a physics-based surrogate to obtain the main characteristics of a good design in a relatively fast iterative process. The case study of a runner design for a low-head Francis turbine indicates advantages of integrating two derivative-free optimization algorithms with different local- and global search capabilities.