• 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.

• 호남수학학술지
• /
• 제43권3호
• /
• pp.547-559
• /
• 2021

We first define the notions of filter continuous, filter sequentially continuous and filter strongly continuous in the framework of fuzzy normed space (FNS), and then we introduce the notion of filter slowly oscillating sequences in the setting of FNS and shows that this notion is stronger than slowly oscillating sequences. Further, we define the concept of filter slowly oscillating continuous functions, filter Cesàro slowly oscillating sequences as well as some other related notions in the aforementioned space and investigate several related results.

• 대한수학회지
• /
• 제46권4호
• /
• pp.813-826
• /
• 2009

In this paper, we generalize the notions of preserving and strongly preserving maps to arbitrary set based topological categories. Further, we obtain characterizations of each of these concepts as well as interprete analogues and generalizations of theorems of Gerlits at al [20] in the categories of filter and local filter convergence spaces.

• Journal of Astronomy and Space Sciences
• /
• 제22권4호
• /
• pp.501-512
• /
• 2005

본 논문에서는 우주개발 중장기계획에 따라 개발 중인 KSLV-I 우주발사체가 나로 우주센터에서 발사될 경우를 가정하여 확장칼만필터와 Unscented 칼만필터를 통한 실시간 궤적추정 결과를 비교 분석하였다. 이를 위해 가상의 KSLV-I과 관측 레이더 3기를 바탕으로 실제 궤적을 생성하였으며, 관측자료는 실제 궤적에 관측오차를 고려하여 생성하였다. 이에 대해 초기 추정오차가 작은 경우와 큰 경우로 구분하고, 관측주기가 20Hz와 10Hz인 경우로 나누어 각각 두 필터를 적용해서 성능을 비교하였다. Unscented 칼만필터는 확장칼만필터보다 큰 초기 추정오차에 대해 수렴이 빠르고 정확도가 높으며, 느린 관측주기에도 우수한 성능을 보이는 것을 확인하였다.

• International Journal of Control, Automation, and Systems
• /
• 제4권5호
• /
• pp.645-652
• /
• 2006

In this paper a new type of filter, called the $H_2/H_{\infty}$ FIR filter, is proposed for discrete-time state space signal models. The proposed filter requires linearity, unbiased property, FIR structure, and independence of the initial state information in addition to the performance criteria in both $H_2$ and $H_{infty}$ sense. It is shown that $H_2,\;H_{\infty}$, and $H_2/H_{\infty}$ FIR filter design problems can be converted into convex programming problems via linear matrix inequalities (LMIs) with a linear equality constraint. Simulation studies illustrate that the proposed FIR filter is more robust against temporary uncertainties and has faster convergence than the conventional IIR filters.

• 대한수학회지
• /
• 제44권4호
• /
• pp.845-854
• /
• 2007

In this paper we investigate relationships between closure-type and convergence-type properties of hyperspaces over a space X and covering properties of X.

• Journal of applied mathematics & informatics
• /
• 제40권1_2호
• /
• pp.327-339
• /
• 2022

In this paper, we introduce arithmetic ${\mathcal{I}}$-statistically convergent sequence space $A{\mathcal{I}}SC$, ${\mathcal{I}}$-lacunary arithmetic statistically convergent sequence space $A{\mathcal{I}}SC_{\theta}$, strongly ${\mathcal{I}}$-lacunary arithmetic convergent sequence space $AN_{\theta}[{\mathcal{I}}]$ and prove some inclusion relations between these spaces. Futhermore, we give ${\mathcal{I}}$-lacunary arithmetic statistical continuity. Finally, we define ${\mathcal{I}}$-Cesàro arithmetic summability, strongly ${\mathcal{I}}$-Cesàro arithmetic summability. Also, we investigate the relationship between the concepts of strongly ${\mathcal{I}}$-Cesàro arithmetic summability, strongly ${\mathcal{I}}$-lacunary arithmetic summability and arithmetic ${\mathcal{I}}$ -statistically convergence.

• 융합신호처리학회논문지
• /
• 제19권4호
• /
• pp.153-161
• /
• 2018

본 논문은 GPU를 이용한 소프트웨어(SW) 디지털 필터의 성능개선에 대해 기술한다. 기존에 개발한 SW 디지털 필터는 CPU 기반에서 동작하여 속도가 느린 문제점이 있었는데, EAVN 관측데이터의 디지털 필터링을 위해 GPU를 도입하여 연산속도를 개선하였고, 필터링을 통하여 다른 관측국과의 데이터 처리가 가능하도록 하였다. SW 디지털 필터의 연산속도를 개선하기 위해 Tensor Core가 내장된 NVIDIA Titan V GPU 보드를 사용하였으며, 2Gbps (512 MHz BW, 1-IF)의 95초 관측데이터를 필터링하는데 관측시간의 약 1.1배, 1Gbps (16MHz BW, 16-IF)로 필터링하는데 약 0.78배 처리속도를 각각 달성하였다. 또한 KVN으로 1, 2Gbps 동시관측한 데이터에 대해 2Gbps 데이터를 디지털 필터링하여 기존 1Gbps와 비교한 결과, 교차전력스펙트럼, 위상, SNR 등이 유사한 값을 얻어 본 연구에서 개발한 SW 디지털 필터를 활용한 데이터 처리와 분석을 수행하는데 유효함을 확인하였다. 향후에는 여러 개의 GPU 보드를 사용하기 위한 소스 코드의 분산처리 최적화를 수행할 경우 실시간으로 관측데이터를 필터링할 수 있을 것으로 기대된다.

• 한국멀티미디어학회논문지
• /
• 제24권2호
• /
• pp.178-185
• /
• 2021

Three-dimensional high-resolution magnetic resonance imaging (MRI) provides fine-level anatomical information for disease diagnosis. However, there is a limitation in obtaining high resolution due to the long scan time for wide spatial coverage. Therefore, in order to obtain a clear high-resolution(HR) image in a wide spatial coverage, a super-resolution technology that converts a low-resolution(LR) MRI image into a high-resolution is required. In this paper, we propose a super-resolution technique through filter learning based on information on the surrounding gradient information in 3D space from 3D MRI images. In the learning step, the gradient features of each voxel are computed through eigen-decomposition from 3D patch. Based on these features, we get the learned filters that minimize the difference of intensity between pairs of LR and HR images for similar features. In test step, the gradient feature of the patch is obtained for each voxel, and the filter is applied by selecting a filter corresponding to the feature closest to it. As a result of learning 100 T1 brain MRI images of HCP which is publicly opened, we showed that the performance improved by up to about 11% compared to the traditional interpolation method.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2014년도 추계학술발표대회
• /
• pp.520-522
• /
• 2014

Due to the availability of easy-to-use and powerful image editing tools, the authentication of digital images cannot be taken for granted and it gives rise to non-intrusive forgery detection problem because all imaging devices do not embed watermark. Forgery detection plays an important role in this case. In this paper, an effective framework for passive-blind method for copy-move image forgery detection is proposed, based on Gabor filter which is robust to illumination, rotation invariant, robust to scale. For the detection, the suspicious image is selected and Gabor wavelet is applied from whole scale space and whole direction space. We will extract the mean and the standard deviation as the texture features and feature vectors. Finally, a distance is calculated between two textures feature vectors to determine the forgery, and the decision will be made based on that result.