• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.26-26
• /
• 2023

General circulation models (GCMs) are widely used in hydrological prediction, however their coarse grids make them unsuitable for regional analysis, therefore a downscaling method is required to utilize them in hydrological assessment. As one of the downscaling methods, convolutional neural network (CNN)-based downscaling has been proposed in recent years. The aim of this study is to generate the process of dynamic downscaling using CNNs by applying GCM output as input and RCM output as label data output. Prediction accuracy is compared between different input datasets, and model structures. Several input datasets with key atmospheric variables such as precipitation, temperature, and humidity were tested with two different formats; one is two-dimensional data and the other one is three-dimensional data. And in the model structure, the hyperparameters were tested to check the effect on model accuracy. The results of the experiments on the input dataset showed that the accuracy was higher for the input dataset without precipitation than with precipitation. The results of the experiments on the model structure showed that substantially increasing the number of convolutions resulted in higher accuracy, however increasing the size of the receptive field did not necessarily lead to higher accuracy. Though further investigation is required for the application, this paper can contribute to the development of efficient downscaling method with CNNs.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• pp.29-32
• /
• 2020

Due to the lack of improper image acquisition process, noise induction is an inevitable step. As a result, objective image quality assessment (IQA) plays an important role in estimating the visual quality of noisy image. Plenty of IQA methods have been proposed including traditional signal processing based methods as well as current deep learning based methods where the later one shows promising performance due to their complex representation ability. The deep learning based methods consists of several convolution layers and down sampling layers for feature extraction and fully connected layers for regression. Usually, the down sampling is performed by using max-pooling layer after each convolutional block. We reveal that this max-pooling causes information loss despite of knowing their importance. Consequently, we propose a better IQA method that replaces the max-pooling layers with strided convolutions to down sample the feature space and since the strided convolution layers have learnable parameters, they preserve optimal features and discard redundant information, thereby improve the prediction accuracy. The experimental results verify the effectiveness of the proposed method.

• Korean Journal of Artificial Intelligence
• /
• v.11 no.4
• /
• pp.1-8
• /
• 2023

In this paper, we propose a black ice detection platform framework using Convolutional Neural Networks (CNNs). To overcome black ice problem, we introduce a real-time based early warning platform using CNN-based architecture, and furthermore, in order to enhance the accuracy of black ice detection, we apply a multi-scale dilation convolution feature fusion (MsDC-FF) technique. Then, we establish a specialized experimental platform by using a comprehensive dataset of thermal road black ice images for a training and evaluation purpose. Experimental results of a real-time black ice detection platform show the better performance of our proposed network model compared to conventional image segmentation models. Our proposed platform have achieved real-time segmentation of road black ice areas by deploying a road black ice area segmentation network on the edge device Jetson Nano devices. This approach in parallel using multi-scale dilated convolutions with different dilation rates had faster segmentation speeds due to its smaller model parameters. The proposed MsCD-FF Net(2) model had the fastest segmentation speed at 5.53 frame per second (FPS). Thereby encouraging safe driving for motorists and providing decision support for road surface management in the road traffic monitoring department.

• Journal of the Chungcheong Mathematical Society
• /
• v.26 no.2
• /
• pp.323-342
• /
• 2013

Let $C[0,t]$ denote the function space of all real-valued continuous paths on $[0,t]$. Define $Xn:C[0,t]{\rightarrow}\mathbb{R}^{n+1}$ and $X_{n+1}:C[0,t]{\rightarrow}\mathbb{R}^{n+2}$ by $X_n(x)=(x(t_0),x(t_1),{\cdots},x(t_n))$ and $X_{n+1}(x)=(x(t_0),x(t_1),{\cdots},x(t_n),x(t_{n+1}))$, where $0=t_0$ < $t_1$ < ${\cdots}$ < $t_n$ < $t_{n+1}=t$. In the present paper, using simple formulas for the conditional expectations with the conditioning functions $X_n$ and $X_{n+1}$, we evaluate the $L_p(1{\leq}p{\leq}{\infty})$-analytic conditional Fourier-Feynman transforms and the conditional convolution products of the functions which have the form $${\int}_{L_2[0,t]}{{\exp}\{i(v,x)\}d{\sigma}(v)}{{\int}_{\mathbb{R}^r}}\;{\exp}\{i{\sum_{j=1}^{r}z_j(v_j,x)\}dp(z_1,{\cdots},z_r)$$ for $x{\in}C[0,t]$, where $\{v_1,{\cdots},v_r\}$ is an orthonormal subset of $L_2[0,t]$ and ${\sigma}$ and ${\rho}$ are the complex Borel measures of bounded variations on $L_2[0,t]$ and $\mathbb{R}^r$, respectively. We then investigate the inverse transforms of the function with their relationships and finally prove that the analytic conditional Fourier-Feynman transforms of the conditional convolution products for the functions, can be expressed in terms of the products of the conditional Fourier-Feynman transforms of each function.

• Journal of the Korean Mathematical Society
• /
• v.50 no.5
• /
• pp.1105-1127
• /
• 2013

Let C[0, $t$] denote the function space of real-valued continuous paths on [0, $t$]. Define $X_n\;:\;C[0,t]{\rightarrow}\mathbb{R}^{n+1}$ and $X_{n+1}\;:\;C[0,t]{\rightarrow}\mathbb{R}^{n+2}$ by $X_n(x)=(x(t_0),x(t_1),{\ldots},x(t_n))$ and $X_{n+1}(x)=(x(t_0),x(t_1),{\ldots},x(t_n),x(t_{n+1}))$, respectively, where $0=t_0 <; t_1 <{\ldots} < t_n < t_{n+1}=t$. In the present paper, using simple formulas for the conditional expectations with the conditioning functions $X_n$ and $X_{n+1}$, we evaluate the $L_p(1{\leq}p{\leq}{\infty})$-analytic conditional Fourier-Feynman transforms and the conditional convolution products of the functions, which have the form $fr((v_1,x),{\ldots},(v_r,x)){\int}_{L_2}_{[0,t]}\exp\{i(v,x)\}d{\sigma}(v)$ for $x{\in}C[0,t]$, where $\{v_1,{\ldots},v_r\}$ is an orthonormal subset of $L_2[0,t]$, $f_r{\in}L_p(\mathbb{R}^r)$, and ${\sigma}$ is the complex Borel measure of bounded variation on $L_2[0,t]$. We then investigate the inverse conditional Fourier-Feynman transforms of the function and prove that the analytic conditional Fourier-Feynman transforms of the conditional convolution products for the functions can be expressed by the products of the analytic conditional Fourier-Feynman transform of each function.

• Economic and Environmental Geology
• /
• v.27 no.1
• /
• pp.41-47
• /
• 1994

A fast Hankel transform (FHT) algorithm (Anderson, 1982) is applied to numerical evaluation of many Green's tensor integrals encountered in three-dimensional electromagnetic modeling using integral equations. Efficient computation of Hankel transforms is obtained by a combination of related and lagged convolutions which are available in the FHT. We express Green's tensor integrals for a layered half-space, and rewrite those to a form of related functions so that the FHT can be applied in an efficient manner. By use of the FHT, a complete or full matrix of the related Hankel transform can be rapidly and accurately calculated for about the same computation time as would be required for a single direct convolution. Computing time for a five-layer half-space shows that the FHT is about 117 and 4 times faster than conventional direct and multiple lagged convolution methods, respectively.

• The Korean Journal of Applied Statistics
• /
• v.22 no.5
• /
• pp.937-942
• /
• 2009

In this paper, a continuous-time risk process in an insurance business is considered, where the premium rate is constant and the claim process forms a compound Poisson process. We say that a ruin occurs if the surplus of the risk process becomes negative. It is practically impossible to calculate analytically the ruin probability because the theoretical formula of the ruin probability contains the recursive convolutions and infinite sum. Hence, many authors have suggested approximation formulas of the ruin probability. We introduce a new approximation formula of the ruin probability which extends the well-known De Vylder's and exponential approximation formulas. We compare our approximation formula with the existing ones and show numerically that our approximation formula gives closer values to the true ruin probability in most cases.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.14 no.1
• /
• pp.1-6
• /
• 2014

Recently, orthogonal frequency-division multiplexing(OFDM), with clusters of subcarriers allocated to different subscribers(often referred to as OFDMA), has gained much attention for its ability in enabling multiple-access wireless multimedia communications. In such systems, carrier frequency offsets (CFOs) can destroy the orthogonality among subcarriers. And the mismatch in sampling frequencies between transmitter and receiver can lead to serious degradation due to the loss of orthogonality between the subcarriers. As a result, multiuser interference (MUI) along with significant performance degradation can be induced. In this paper, we present a scheme to compensate for the SFOs and CFOs at the base station of an OFDMA system. A novel sampling frequency offset estimation algorithm is proposed, which is based on the repetition of a symbol at the communication start-up. Then, circular convolutions are employed to generate the interference after the discrete Fourier transform processing, which is then removed from the original received signal to increase the signal to interference power ratio(SIR). Simulation result shows that the proposed scheme can improve system performance.

• Journal of Advanced Marine Engineering and Technology
• /
• v.40 no.7
• /
• pp.587-592
• /
• 2016

Bellows for LNG carriers must be corrosion resistant in order to operate in seawater environments. They must also have long fatigue lives in order to withstand the expansion and contraction caused by large temperature changes and continuous vibration in extreme environments. In order to incorporate these properties into bellow design, it is important to use materials that are resistant to cold brittleness and corrosion, and maintain their optimized forming condition. The design conditions and forming methods used for bellows must be optimized in order to incorporate these characteristics. In this study, finite element analysis was used to develop cryogenic bellows, which have good mechanical strength and reliability. In addition, two different forming methods (mechanical and hydroforming) were used to design and produce bellows, in order to derive their forming condition. The height, thickness, and hardness of the convolutions of bellows produced by each method were measured and compared with each other. The results confirmed that the two forming methods produced bellows with different mechanical properties.

• Journal of the Institute of Convergence Signal Processing
• /
• v.5 no.2
• /
• pp.159-164
• /
• 2004

This paper proposes a new digital filter implementation which adopts an identical structure at both behavioral and logic level in top-down design. This methodology is based on the observation that multiplication is a form of convolution and carrying, and therefore multiplication is implemented with the same structure as that of a convolution in a recursive manner at the logic level. In order to demonstrate a recursive structure-based FIR digital filter, we select L-tap transposed and systolic FIR filters, and implement them to have a single structure. The proposed filter design becomes regular and modular because of the recursive adoption of a single structure for convolutions, and is very compact in that it needs only two 1-bit I/O ports in addition to significant improvement on hardware complexity without time penalty on the output sequence.