• Journal of the Korean Data and Information Science Society
• /
• v.17 no.4
• /
• pp.1289-1297
• /
• 2006

Before other statistical data analysis the preprocessing steps should be performed adequately to have meaningful results. These steps include processes such as baseline correction, normalization, denoising, and multiple alignment. In this paper an algorithm for baseline correction is proposed with using the piecewise cubic Hermite interpolation with block-selected points and local minima after denoising for SELDI or MALDI mass spectrometry data.

• 제어로봇시스템학회:학술대회논문집
• /
• 2002.10a
• /
• pp.45.6-45
• /
• 2002

In this paper, a novel optimization algorithm which searches for the local minima of a given cost function is proposed using the familiar property of a binary string, and is applied to the parametric identification of a continuous-time state equation by the estimation of system parameters as well as initial state values. A simple electrical circuit severs as an example, whose precise identification results show the superiority of the proposed algorithm.

• Proceedings of the IEEK Conference
• /
• 1999.06a
• /
• pp.460-463
• /
• 1999

In this paper, we proposed an block-based evolvable neural network(BENN). The BENN can optimize it＇s structure and weights simultaneously. It can be easily implemented by FPGA whose connection and internal functionality can be reconfigured. To solve the local minima problem that is caused gradient descent learning algorithm, genetic algorithms are applied for optimizing the proposed evolvable neural network model.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.5
• /
• pp.661-670
• /
• 1995

This paper presents a fast massively parallel automatic test pattern generator for digital combinational logic circuits using neural networks. Automatic test pattern generation neural network(ATPGNN) evolves its state to a stable local minima by exchanging messages among neural network modules. In preprocessing phase, we calculate the essential assignments for the stuck-at faults in fault list by adopting dominator concept. It makes more neurons be fixed and the system speed up. Consequently. fast test pattern generation is achieved. Test patterns for stuck-open faults are generated through getting initialization patterns for the obtained stuck-at faults in the corresponding ATPGNN.

• Proceedings of the Korean Information Science Society Conference
• /
• 1998.10c
• /
• pp.476-478
• /
• 1998

본 논문에서는 영상의 밝기 정보로부터 물체의 표면 형상을 재구성하는 새로운 접근 방법을 제시한다. 이미지 모델은 기존의 Lambertian surface model에 거리 요소를 포함시켜 보다 현실과 비슷한 제약 조건을 주고, 국지 해(local minima)에 빠지기 쉬운 기존의 iteration 방법을 탈피하기 위해 유전자 알고리즘(genetic algorithm)을 도입한다. 표면의 깊이 정보를 이산여현변환(discrete cosine transform)하고 이 DCT 공간상에서 유전자 알고리즘을 적용함으로써 큰 형상을 먼저 결정한 후 미세한 형상을 찾아내는 계층적인 표면 형상의 재구성이 가능하도록 하였으며 간단한 실험으로 그 타당성을 보인다.

• Proceedings of the Korea Contents Association Conference
• /
• 2005.05a
• /
• pp.362-368
• /
• 2005

This paper is related to the fast block matching algorithm, especially a method for expanding the search pattern using the second local matching point in the adaptive hexagonal search. To reduce the local minima problem in fast motion estimation, the proposed method expands the search pattern by adding new searching points selected by using the second local matching point to conventional search pattern formed by the first local matching point in the adaptive hexagonal search. According to estimating the motion vector by applying block matching algorithm based on hexagonal search to the expanded search pattern, the proposed method can effectively carry out fast motion estimation to improve the performance in terms of compensated image quality.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.43 no.6 s.312
• /
• pp.57-64
• /
• 2006

Motion estimation plays an important role for the compression of video signals. The proposed method utilizes an adaptive threshold and characteristics of a distribution of SAD (sum of absolute difference). Generally, the more complex the SAD distribution is, the larger SAD value tends to be. This proposed algorithm tries to reduce the search points in a simple distribution but increase them in a complex distribution to avoid local minima. A macro block is divided into 9 areas. One of them chosen using spatio-temporal correlation is called the primary area and the others are called the secondary area that will be searched to avoid local minima. The proposed algorithm decides if just one area (the primary area or the secondary area) will be enough to be searched or both areas should be searched, using adaptive threshold. Compared with famous motion estimation algorithms, the simulation result shows that the searching points per macro block and MSE decreases about 16.4% and 32.83 respectively on the average.

• Journal of Korea Multimedia Society
• /
• v.17 no.2
• /
• pp.143-151
• /
• 2014

In images degraded by bad weather, edges representing object boundaries become weak and faint. In this paper we present an image enhancement method, which increases image visibility by making edges as clear as possible. First, we choose edge candidate regions by finding local maxima and minima in an image intensity field, and then build a histogram using image intensities of pixels located at the two sides of candidate edges. Second, we decompose this histogram into multiple modes, which are determined by local minima in the histogram. Once modes are computed, we find modes connected by edges in the image intensity field and build link chains of connected modes. Finally we choose the longest link chain of modes and make the distances between every connected modes as large as possible. The darkest mode and the brightest mode should be within the image intensity range. This stretch makes edges clear and increases image visibility. Experiments show that the proposed method real-time enhances images degraded by bad weather as good as well known time-consuming methods.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.9 no.4
• /
• pp.883-889
• /
• 2005

Error backpropagation algorithm of multilayer perceptron may result in local-minima because of the insufficient nodes in the hidden layer, inadequate momentum set-up, and initial weights. In this paper, we proposed the ART-1 based fuzzy supervised learning algorithm which is composed of ART-1 and fuzzy single layer supervised learning algorithm. The Proposed fuzzy supervised learning algorithm using self-generation method applied not only ART-1 to creation of nodes from the input layer to the hidden layer, but also the winer-take-all method, modifying stored patterns according to specific patterns. to adjustment of weights. We have applied the proposed learning method to the problem of recognizing a resident registration number in resident cards. Our experimental result showed that the possibility of local-minima was decreased and the teaming speed and the paralysis were improved more than the conventional error backpropagation algorithm.

• Proceedings of the Korean Biophysical Society Conference
• /
• 2003.06a
• /
• pp.75-75
• /
• 2003

We report here the results on N-acetyl-N'-dimethylamide of proline (Ac-Pro-NM $e_2$) calculated using the ab initio molecular orbital method with the self-consistent reaction field (SCRF) theory at the HF level with the 6-31+G(d) basis set to investigate the conformational preference of polyproline depending on the cis/trans peptide bonds and down/up puckerings along the backbone torsion angle $\square$ in the gas phase, chloroform, and water. In the gas phase, Ac-Pro-NM $e_2$ has seven local minima of tFd, tFu, cFd, cFu, cAu, tAu, and cAd conformations. In particular, polyproline conformations tFd, tFu, cFd, and cFu are found to be more stable than $\square$-helical conformations cAu, tAu, and cAd. In contrast, Ac-Pro-NHMe has seven local minima of tCd, tCu, cBd, cAu, tAu, cFd, and cFu conformations. Conformations tCd and tCu are found to be most stable, which is ascribed to the intramolecular hydrogen bond between C=O of acetyl group and $N^{~}$ H of N'-methyl amide group. The stability of the cFd conformation (i.e., the polyproline I structure) in chloroform is somewhat increased, relative to that in water, although tFd and tFu conformations (i.e., the polyproline II structure) are dominate both in chloroform and water. The population of backbone conformations feasible in chloroform and water is consistent with the experiments. This work is supported by a Korea Research Foundation Grant (KRF-2002-041-C00129).