• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.11
• /
• pp.4065-4083
• /
• 2021

Super-resolution can improve the clarity of low-resolution (LR) images, which can increase the accuracy of high-level compute vision tasks. Portable devices have low computing power and storage performance. Large-scale neural network super-resolution methods are not suitable for portable devices. In order to save the computational cost and the number of parameters, Lightweight image processing method can improve the processing speed of portable devices. Therefore, we propose the Enhanced Information Multiple Distillation Network (EIMDN) to adapt lower delay and cost. The EIMDN takes feedback mechanism as the framework and obtains low level features through high level features. Further, we replace the feature extraction convolution operation in Information Multiple Distillation Block (IMDB), with Ghost module, and propose the Enhanced Information Multiple Distillation Block (EIMDB) to reduce the amount of calculation and the number of parameters. Finally, coordinate attention (CA) is used at the end of IMDB and EIMDB to enhance the important information extraction from Spaces and channels. Experimental results show that our proposed can achieve convergence faster with fewer parameters and computation, compared with other lightweight super-resolution methods. Under the condition of higher peak signal-to-noise ratio (PSNR) and higher structural similarity (SSIM), the performance of network reconstruction image texture and target contour is significantly improved.

• Journal of Korean Neurosurgical Society
• /
• v.67 no.1
• /
• pp.94-102
• /
• 2024

Objective : The spontaneous intracerebral hemorrhage (ICH) remains a significant cause of mortality and morbidity throughout the world. The purpose of this retrospective study is to develop multiple models for predicting ICH outcomes using machine learning (ML). Methods : Between January 2014 and October 2021, we included ICH patients identified by computed tomography or magnetic resonance imaging and treated with surgery. At the 6-month check-up, outcomes were assessed using the modified Rankin Scale. In this study, four ML models, including Support Vector Machine (SVM), Decision Tree C5.0, Artificial Neural Network, Logistic Regression were used to build ICH prediction models. In order to evaluate the reliability and the ML models, we calculated the area under the receiver operating characteristic curve (AUC), specificity, sensitivity, accuracy, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR). Results : We identified 71 patients who had favorable outcomes and 156 who had unfavorable outcomes. The results showed that the SVM model achieved the best comprehensive prediction efficiency. For the SVM model, the AUC, accuracy, specificity, sensitivity, PLR, NLR, and DOR were 0.91, 0.92, 0.92, 0.93, 11.63, 0.076, and 153.03, respectively. For the SVM model, we found the importance value of time to operating room (TOR) was higher significantly than other variables. Conclusion : The analysis of clinical reliability showed that the SVM model achieved the best comprehensive prediction efficiency and the importance value of TOR was higher significantly than other variables.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.9 no.1
• /
• pp.30-38
• /
• 1989

Since the nuclear power plants in Korea have been operated in 1979, the nondestructive testing (NDT) of pressure vessels and/or piping welds plays an important role for maintaining the safety and integrity of the plants. Ultrasonic method is superior to the other NDT method in the viewpoint of the detectability of small flaw and accuracy to determine the locations, sizes, orientations, and shapes. As the service time of the nuclear power plants is increased, the radiation level from the components is getting higher. In order to get more quantitative and reliable results and secure the inspector from the exposure to high radiation level, automation of the ultrasonic equipments has been one of the important research and development(R & D) subject. In this research, it was attempted to visualize the shape of flaws presented inside the specimen using a Modified C-Scan technique. In order to develope Modified C-Scan technique, an automatic ultrasonic scanner and a module to control the scanner were designed and fabricated. IBM-PC/XT was interfaced to the module to control the scanner. Analog signals from the SONIC MARK II were digitized by Analog-Digital Converter(ADC 0800) for Modified C-Scan display. A computer program has been developed and has capability of automatic data acquisition and processing from the digital data, which consist of maximum amplitudes in each gate range and locations. The data from Modified C-Scan results was compared with shape from artificial defects using the developed system. Focal length of focused transducer was measured. The automatic ultrasonic equipment developed through this study is essential for more accurate, reliable, and repeatable ultrasonic experiments. If the scanner are modified to meet to appropriate purposes, it can be applied to automation of ultrasonic examination of nuclear power plants and helpful to the research on ultrasonic characterization of the materials.

• Journal of Ocean Engineering and Technology
• /
• v.25 no.2
• /
• pp.7-14
• /
• 2011

Numerical simulations were performed for the evaluation of wave and current loads on a fixed cylindrical substructure model for an ocean wind turbine using the ANSYS-CFX package. The numerical wave tank was actualized by specifying the velocity at the inlet and applying momentum loss as a wave damper at the end of the wave tank. The Volume-Of-Fluid (VOF) scheme was adopted to capture the air-water interface. An accuracy validation of the numerical wave tank with a truncated vertical circular cylinder was accomplished by comparing the CFD results with Morison's formula, experimental results, and potential flow solutions using the higher-order boundary element method (HOBEM). A parametric study was carried out by alternately varying the length and amplitude of the wave. As a meaningful engineering application, in the present study, three kinds of conditions were considered, i.e., cases with current, waves, and a combination of current and progressive waves, passing through a cylindrical substructure model. It was found that the CFD results showed reasonable agreement with the results of the HOBEM and Morison's formula when only progressive waves were considered. However, when a current was included, CFD gave a smaller load than Morison's formula.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.6 no.2
• /
• pp.12-20
• /
• 1998

Exter ballistics of a typical high-speed projectile is studied through a flow-visualization experiment and an unstructured grid Navier-Srokes computation. Experiment produced a schlieren photograph that adequately shows the characteristic features of this complex flow, namely two kinds of oblique cone shocks and turbulent wake developing into the downstream. A hybrid scheme of finite volume-element method is used to simulate the compressible Reynolds-Averaged Navier-Stok- es solution on unstructured grids. Osher's approximate Riemann solver is used to discretize the cinvection term. Higher-order spatial accuracy is obtained by MUSCL extension and van Albada ty- pe flux limiter is used to stabilize the numerical oscillation near the solution discontinuity. Accurate Gakerkin method is used to discretize the viscous term. Explict fourth-order Runge-Kutta method is used for the time-stepping, which simplifies the application of MUSCL extension. A two-layer k-$\varepsilon$ turbulence model is used to simulate the turbulent wakes accurately. Axisymmetric folw and two-dimensional flow with an angle of attack have been computed. Grid-dependency is also checked by carrying out the computation with doubled meshes. 2-D calculation shows that effect of angle of attack on the flow field is negligible. Axi-symmetric results of the computation agrees well with the flow visualization. Primary oblique shock is represented within 2-3 meshes in numerical results, and the varicose mode of the vortex shedding is clearly captured in the turbulent wake region.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.49 no.7
• /
• pp.378-388
• /
• 2000

In this paper, we propose Fuzzy Polynomial Neural Networks(FPNN) based on Polynomial Neural Networks(PNN) and Fuzzy Neural Networks(FNN) for model identification of complex and nonlinear systems. The proposed FPNN is generated from the mutually combined structure of both FNN and PNN. The one and the other are considered as the premise part and consequence part of FPNN structure respectively. As the consequence part of FPNN, PNN is based on Group Method of Data Handling(GMDH) method and its structure is similar to Neural Networks. But the structure of PNN is not fixed like in conventional Neural Networks and self-organizing networks that can be generated. FPNN is available effectively for multi-input variables and high-order polynomial according to the combination of FNN with PNN. Accordingly it is possible to consider the nonlinearity characteristics of process and to get better output performance with superb predictive ability. As the premise part of FPNN, FNN uses both the simplified fuzzy inference as fuzzy inference method and error back-propagation algorithm as learning rule. The parameters such as parameters of membership functions, learning rates and momentum coefficients are adjusted using genetic algorithms. And we use two kinds of FNN structure according to the division method of fuzzy space of input variables. One is basic FNN structure and uses fuzzy input space divided by each separated input variable, the other is modified FNN structure and uses fuzzy input space divided by mutually combined input variables. In order to evaluate the performance of proposed models, we use the nonlinear function and traffic route choice process. The results show that the proposed FPNN can produce the model with higher accuracy and more robustness than any other method presented previously. And also performance index related to the approximation and prediction capabilities of model is evaluated and discussed.

• Journal of the Earthquake Engineering Society of Korea
• /
• v.9 no.3 s.43
• /
• pp.33-42
• /
• 2005

Frequency domain Volterra model is applied to nonlinear parameter identification procedure for dynamic systems modeled by nonlinear function. The frequency domain Volterra kernels, which correspond io linear, quadratic, and cubic transfer functions in lime domain, are incorporated in nonlinear parametric identification procedure. The nonlinear transfer functions, which can be derived from the Volterra series representation of the nonlinear differential equation of the system by Schetzen's method(1980), are directly used for modeling input output relation. The error is defined by the difference between the observed output and the estimated output which is calculated by substituting the observed input to nonlinear frequency domain model. The system parameters are searched by minimizing the error. Volterra model guarantees enough accuracy and convergence and the estimated coefficients have a good agreement with their actual values not only in the linear frequency region but also in the legion where the $2^{nd}\;or\;3^{rd}$ order nonlinearity is dominant.

• Physical Therapy Korea
• /
• v.19 no.1
• /
• pp.10-18
• /
• 2012

The purpose of this study was to determine the effects of weight shift training with joint mobilization on the ankle joint passive range of motion (PROM), balance capacity and gait velocity in hemiplegic patients. Fourteen subjects were randomly assigned to either the experimental group (EG) or the control group (CG), with seven subjects in each group. The EG received weight shift training with joint mobilization in the paretic leg's subtalar joint in order to increase ankle dorsiflexion. The CG received general physical therapy training. Both groups received training five times a week over a period of two consecutive weeks. The figures for PROM of ankle dorsiflexion on the paretic leg, the functional reach test (FRT), the timed up and go (TUG) test, and gait velocity were recorded both before and after the training sessions for both groups. The EG's results in gait velocity, the FRT and the TUG test improved after training (p<.05). The PROM of ankle dorsiflexion improved both in the EG and the CG (p<.05), the EG demonstrated a significantly higher increase (p<.05) than that of the CG. The results of this study suggest that increased joint mobilization positively affects balance and gait velocity of hemiplegic patients. Further studies with a greater sample size are necessary in order further prove the accuracy of the results of this study.

• The Journal of the Korea Contents Association
• /
• v.13 no.11
• /
• pp.37-47
• /
• 2013

Recently, lots of research results on program comparison have been reported since the code theft become frequent as the increase of code mobility. This paper proposes a plagiarism detection method using class structures. The proposed method constructs a graph representing the referential relationship between the member variables and the methods. This relationship is shown as a bipartite graph and the test for graph isomorphism is applied on the set of graphs to measure the similarity of the programs. In order to measure the effectiveness of this method, an experiment was conducted on the test set, the set of Java source codes submitted as solutions for the programming assignments in Object-Oriented Programming course of Pusan National University in 2012. In order to evaluate the accuracy of the proposed method, the F-measure is compared to those of JPlag and Stigmata. According to the experimental result, the F-measure of the proposed method is higher than those of JPlag and Stigmata by 0.17 and 0.34, respectively.

• Journal of the Korean Society for Precision Engineering
• /
• v.25 no.4
• /
• pp.134-139
• /
• 2008

A spirally arrayed nano-pattern is designed as a model pattern for the next generation optical storage media. The pattern consists off types of embossed rectangular dot, which are 50nm, 100nm, 150nm and 200nm in length and 50nm in width. The height of the dot is designed to be 50nm. The pitch of the spiral track of the pattern is 100nm. A ER(Electron resist) master for this pattern is fabricated by e-beam lithography process. The ER is first spin-coated to be 50nm thick on a Si wafer and then the model pattern is written on the coated ER layer by e-beam. After developing this pattern written wafer in the solution, a ER pattern master is fabricated. The most conventional e-beam machine can write patterns in orthogonal way, so we made our own pattern generator which can write the pattern in circular or spiral way. This program generates the patterns to be compatible with the e-beam machine from Raith(Raith 150). To fabricate 50nm pattern master precisely, a series of experiments were done including the design compensation for the pattern size, optimization of the dose, acceleration voltage, aperture size and developing. Through these experiments, we conclude that the higher accelerating voltages and smaller aperture size are better for mastering the nano pattern which is in order of 50nm. With the optimized e-beam lithography process, a spiral arrayed 50nm pattern master adopting PMMA resist was fabricated to have dimensional accuracy over 95% compared to the designed. Using this pattern master, a metal pattern stamp will be fabricated by Ni electro plating for injection molding of the patterned plastic substrate.