• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.32 no.6
• /
• pp.561-568
• /
• 2020

Wave measurements using X-band radar have many advantages compared to other wave gauges including wave-rider buoy, P-u-v gauge and Acoustic Doppler Current Profiler (ADCP), etc.. For example, radar system has no risk of loss/damage in bad weather conditions, low maintenance cost, and provides spatial distribution of waves from deep to shallow water. This paper presents new methods for estimating significant wave heights of X-band marine radar images using Artificial Neural Network (ANN). We compared the time series of estimated significant wave heights (Hs) using various estimation methods, such as signal-to-noise ratio (${\sqrt{SNR}}$), both and ${\sqrt{SNR}}$ the peak period (TP), and ANN with 3 parameters (${\sqrt{SNR}}$, TP, and Rval > k). The estimated significant wave heights of the X-band images were compared with wave measurement using ADCP(AWC: Acoustic Wave and Current Profiler) at Hujeong Beach, Uljin, Korea. Estimation of Hs using ANN with 3 parameters (${\sqrt{SNR}}$, TP, and Rval > k) yields best result.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.10
• /
• pp.4080-4097
• /
• 2020

In recent years, convolutional neural network (CNN) has become the primary method for face detection. But its shortcomings are obvious, such as expensive calculation, heavy model, etc. This makes CNN difficult to use on the mobile devices which have limited computing and storage capabilities. Therefore, the design of lightweight CNN for face detection is becoming more and more important with the popularity of smartphones and mobile Internet. Based on the CPU real-time face detector FaceBoxes, we propose a multi-task lightweight face detector, which has low computing cost and higher detection precision. First, to improve the detection capability, the squeeze and excitation modules are used to extract attention between channels. Then, the textual and semantic information are extracted by shallow networks and deep networks respectively to get rich features. Finally, the landmark detection module is used to improve the detection performance for small faces and provide landmark data for face alignment. Experiments on AFW, FDDB, PASCAL, and WIDER FACE datasets show that our algorithm has achieved significant improvement in the mean average precision. Especially, on the WIDER FACE hard validation set, our algorithm outperforms the mean average precision of FaceBoxes by 7.2%. For VGA-resolution images, the running speed of our algorithm can reach 23FPS on a CPU device.

• Korean Journal of Soil Science and Fertilizer
• /
• v.39 no.5
• /
• pp.259-267
• /
• 2006

Hydraulic conductivity is the rate of water flux on hydraulic gradient. The van Genuchten Mualem (VGM) model is frequently used for describing unsaturated state of soils, that is composed with the function of soil water potential and soil water content and requests various parameters. This study is to get the value of VGM parameters used Rosetta computer program based on neural network analysis method and to calculate VGM parameters. VGM parameters included Ko(effective saturated hydraulic conductivity), ${\theta}r$(residual soil water content), ${\theta}s$(saturated soil water content), L, n and m. The unsaturated hydraulic conductivity at 10 kPa was calculated by using Rosetta program. Unsaturated hydraulic conductivities of 17 soil series at 1, 3, 5, 7 kPa were also obtained by applying saturated hydraulic conductivity by disk tension infiltrometer based on Gardner and Wooding's equation. Water flow at the water potential of 3 kPa was very low except Namgye, Hagog, Baegsan, Sangju, Seogcheon, Yesan soil series. Unsaturated hydraulic conductivity at 1 kPa showed the highest value for Samgag soil series and was in order of Yesan, Hwabong, Hagog and Baegsan soil series. Those of Gacheon, Seocheon and Ugog soil series were very low. When the value by VGM was compared with the value by disc tension infiltrometer, there was a tendency with exponential function to soils without gravel but there was no tendency to soils including gravel. Conclusively, it would be limited that VGM model for unsaturated hydraulic conductivity analysis applies to Korean agricultural land including gravel and having steep slope, shallow soil depth.

• Journal of Ocean Engineering and Technology
• /
• v.32 no.6
• /
• pp.447-457
• /
• 2018

In this study, a two-dimensional fully nonlinear transient wave numerical tank was developed using a desingularized indirect boundary integral equation method. The desingularized indirect boundary integral equation method is simpler and faster than the conventional boundary element method because special treatment is not required to compute the boundary integral. Numerical simulations were carried out in the time domain using the fourth order Runge-Kutta method. A mixed Eulerian-Lagrangian approach was adapted to reconstruct the free surface at each time step. A numerical damping zone was used to minimize the reflective wave in the downstream region. The interpolating method of a Gaussian radial basis function-type artificial neural network was used to calculate the gradient of the free surface elevation without element connectivity. The desingularized indirect boundary integral equation using an isolated point source and radial basis function has no need for information about the element connectivity and is a meshless method that is numerically more flexible. In order to validate the accuracy of the numerical wave tank based on the desingularized indirect boundary integral equation method and meshless technique, several numerical simulations were carried out. First, a comparison with numerical results according to the type of desingularized source was carried out and confirmed that continuous line sources can be replaced by simply isolated sources. In addition, a propagation simulation of a $2^{nd}$-order Stokes wave was carried out and compared with an analytical solution. Finally, simulations of propagating waves in shallow water and propagating waves over a submerged bar were also carried and compared with published data.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.3
• /
• pp.1100-1118
• /
• 2021

In a wide range of ML applications, the training data contains privacy-sensitive information that should be kept secure. Training the ML systems by privacy-sensitive data makes the ML model inherent to the data. As the structure of the model has been fine-tuned by training data, the model can be abused for accessing the data by the estimation in a reverse process called model inversion attack (MIA). Although, MIA has been applied to shallow neural network models of recognizers in literature and its threat in privacy violation has been approved, in the case of a deep learning (DL) model, its efficiency was under question. It was due to the complexity of a DL model structure, big number of DL model parameters, the huge size of training data, big number of registered users to a DL model and thereof big number of class labels. This research work first analyses the possibility of MIA on a deep learning model of a recognition system, namely a face recognizer. Second, despite the conventional MIA under the white box scenario of having partial access to the users' non-sensitive information in addition to the model structure, the MIA is implemented on a deep face recognition system by just having the model structure and parameters but not any user information. In this aspect, it is under a semi-white box scenario or in other words a gray-box scenario. The experimental results in targeting five registered users of a CNN-based face recognition system approve the possibility of regeneration of users' face images even for a deep model by MIA under a gray box scenario. Although, for some images the evaluation recognition score is low and the generated images are not easily recognizable, but for some other images the score is high and facial features of the targeted identities are observable. The objective and subjective evaluations demonstrate that privacy cyber-attack by MIA on a deep recognition system not only is feasible but also is a serious threat with increasing alert state in the future as there is considerable potential for integration more advanced ML techniques to MIA.