• Structural Engineering and Mechanics
• /
• 제5권2호
• /
• pp.115-126
• /
• 1997

Importance sampling methods have been developed with the aim of reducing the computational costs inherent in Monte Carlo methods. This study proposes a new algorithm called the adaptive kernel method which combines and modifies some of the concepts from adaptive sampling and the simple kernel method to evaluate the structural reliability of time variant problems. The essence of the resulting algorithm is to select an appropriate starting point from which the importance sampling density can be generated efficiently. Numerical results show that the method is unbiased and substantially increases the efficiency over other methods.

• 한국산학기술학회논문지
• /
• 제22권3호
• /
• pp.627-632
• /
• 2021

적응 신호 처리 및 머신 러닝 등에 활용되고 있는 정보 이론적 학습법(ITL, information theoretic learning)은 커널 사이즈(��) 설정이 성능에 큰 영향을 미친다. ITL 기반의 학습법의 하나인 코렌트로피 알고리듬은 충격성 잡음에 강인성과 채널 왜곡 보상 특성을 함께 지니고 있으나 커널 사이즈 선택에 매우 민감하거나 불안정한 특성도 지니고 있다. 이에, 이 논문에서는 기울기 분모에 나타나는 커널 사이즈의 세제곱이 미치는 민감성을 고려하고, 커널 사이즈의 미세 변동에 대한 오차 전력 변화율을 이용하여 커널 사이즈를 적응적으로 갱신하는 방법을 제안하여 코렌트로피 알고리듬에 적용하였다. 제안된 적응 커널 사이즈 추정 방법을 다중 경로 채널과 충격성 잡음 환경에 대해 실험하였다. 제안한 방식은 고정 커널사이즈의 기존 알고리듬에 비해 2배 빠른 수렴 속도를 나타냈고 초기 커널 사이즈 2.0 에서 6.0 에 대해 모두 적절히 수렴하는 능력을 보였다. 이에 초기 커널 사이즈 선택에 큰 여유도를 가지고 성능을 향상시킬 수 있음을 입증하였다.

• 한국산학기술학회논문지
• /
• 제22권5호
• /
• pp.561-566
• /
• 2021

오차분포 추정을 위한 커널 사이즈는 오차확률밀도 사이의 유클리드 거리를 최소화 알고리즘의 가중치 갱신에 적합한 커널 사이즈가 될 수 없다. 이 논문에서는 MED 알고리즘의 수렴 성능 향상을 위해 적응적으로 커널 사이즈를 갱신하는 방법을 제안하였다. 제안한 방식은 MED 학습 알고리즘의 가중치 갱신을 위해 커널 사이즈에 대한 오차분산의 평균변화율을 도입하여 MED의 오차에 대한 평균전력이 감소하는 방향으로 커널 사이즈를 조절하도록 하였다. 제안된 적응 커널 추정법을 무선통신 채널의 왜곡 보상에 적용하여 학습 성능을 실험하고 그 효능을 밝혔다. 오차분산에 비례한 작은 값을 가지는 기존의 오차분포 추정 위한 최적 커널 사이즈와 달리, 제안한 방법에 의한 커널 사이즈는 MED 가중치 수렴을 위한 적절한 커널 사이즈로 수렴함을 보였다. 실험 결과로부터 제안한 방법이 MED 알고리즘의 커널 사이즈 설정에 따른 민감성을 크게 해결한 방법이라고 볼 수 있다.

• Communications for Statistical Applications and Methods
• /
• 제2권1호
• /
• pp.99-111
• /
• 1995

It is shown that the adaptive kernel methods can potentially produce superior density estimates to the fixed one. In using the adaptive estimates, problems pertain to the initial choice of the estimate can be solved by iteration. Also, simultaneous recommended for variety of distributions. Some data-based method for the choice of the parameters are suggested based on simulation study.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1995년도 Proceedings of the Korea Automation Control Conference, 10th (KACC); Seoul, Korea; 23-25 Oct. 1995
• /
• pp.476-479
• /
• 1995

A real time and adaptive method for obtaining Volterra kernels of a nonlinear system by use of pseudorandom M-sequences and correlation technique is proposed. The Volterra kernels are calculated real time and the obtained Volterra kernels becomes more accurate as time goes on. The simulation results show the effectiveness of this method for identifying time-varying nonlinear system.

• Structural Engineering and Mechanics
• /
• 제75권6호
• /
• pp.771-784
• /
• 2020

Reliability analysis techniques combining with various surrogate models have attracted increasing attention because of their accuracy and great efficiency. However, they primarily focus on the structures with continuous response, while very rare researches on the reliability analysis for structures with discontinuous response are carried out. Furthermore, existing adaptive reliability analysis methods based on importance sampling (IS) still have some intractable defects when dealing with small failure probability, and there is no related research on reliability analysis for structures involving discontinuous response and small failure probability. Therefore, this paper proposes a novel reliability analysis method called AGPC-IS for such structures, which combines adaptive Gaussian process classification (GPC) and adaptive-kernel-density-estimation-based IS. In AGPC-IS, an efficient adaptive strategy for design of experiments (DoE), taking into consideration the classification uncertainty, the sampling uniformity and the regional classification accuracy improvement, is developed with the purpose of improving the accuracy of Gaussian process classifier. The adaptive kernel density estimation is introduced for constructing the quasi-optimal density function of IS. In addition, a novel and more precise stopping criterion is also developed from the perspective of the stability of failure probability estimation. The efficiency, superiority and practicability of AGPC-IS are verified by three examples.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 1997년도 가을 학술발표회 논문집
• /
• pp.16-23
• /
• 1997

The Reproducing Kernel Particle Method (RKPM), one of the popular meshless methods, is developed and applied to stress concentration problems. Since the meshless methods require only a set of particles (or nodes) and the description of boundaries in their formulation, the adaptivity can be implemented with much more ease than finite element method. In addition, due to its intrinsic property of multiresolution, the shape function of RKPM provides us a new criterion for adaptivity. Recently, this multiple scale Reproducing Kernel Particle Method and its adaptive procedure have been formulated for large deformation problems by the authors. They are also under development for damage materials and localization problems. In this paper the multiple scale RKPM for linear elasticity is presented and the adaptive procedure is applied to stress concentration problems. Therefore, this work may be regarded as the edition of linear elasticity in the complete framework of multiple scale RKPM and the associated adaptivity.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제9권6호
• /
• pp.2302-2316
• /
• 2015

We propose an efficient method of extracting targets within a region of interest in non-homogeneous infrared images by using a principal component analysis (PCA) plane and adaptive Gaussian kernel. Existing approaches for extracting targets have been limited to using only the intensity values of the pixels in a target region. However, it is difficult to extract the target regions effectively because the intensity values of the target region are mixed with the background intensity values. To overcome this problem, we propose a novel PCA based approach consisting of three steps. In the first step, we apply a PCA technique minimizing the total least-square errors of an IR image. In the second step, we generate a binary image that consists of pixels with higher values than the plane, and then calculate the second derivative of the sum of the square errors (SDSSE). In the final step, an iteration is performed until the convergence criteria is met, including the SDSSE, angle and labeling value. Therefore, a Gaussian kernel is weighted in addition to the PCA plane with the non-removed data from the previous step. Experimental results show that the proposed method achieves better segmentation performance than the existing method.

• 전기학회논문지
• /
• 제56권4호
• /
• pp.795-798
• /
• 2007

In this paper, we propose an image reconstruction method utilizing an optimized adaptive interpolation kernel along with a 2D M-channel perfect reconstruction filter bank (M-ch PR-FB) structure. In particular, the proposed approach leads to construction of a sharper image than a direct conversion, still preserving high frequency components of the original image through the subband processing of the 2D M-ch PR-FB. Finally, the image quality of the proposed approach is demonstrated by comparing with those of the direct methods using conventional interpolation kernels.

• 한국정보통신설비학회:학술대회논문집
• /
• 한국정보통신설비학회 2009년도 정보통신설비 학술대회
• /
• pp.382-386
• /
• 2009

Online or sequential learning is one of the most basic and powerful method to train neuron network, and it has been widely used in disease detection, weather prediction and other realistic classification problem. At present, there are many algorithms in this area, such as MRAN, GAP-RBFN, OS-ELM, SVM and SMC-RBF. Among them, SMC-RBF has the best performance; it has less number of hidden neurons, and best efficiency. However, all the existing algorithms use signal normal distribution as kernel function, which means the output of the kernel function is same at the different direction. In this paper, we use multi-variable normal distribution as kernel function, and derive EKF learning formulas for multi-variable normal distribution kernel function. From the result of the experience, we can deduct that the proposed method has better efficiency performance, and not sensitive to the data sequence.