• International journal of advanced smart convergence
• /
• v.12 no.2
• /
• pp.56-66
• /
• 2023

The goal of deep learning is to extract complex features from multidimensional data use the features to create models that connect input and output. Deep learning is a process of learning nonlinear features and functions from complex data, and the user data that is employed to train deep learning models has become the focus of privacy concerns. Companies that collect user's sensitive personal information, such as users' images and voices, own this data for indefinite period of times. Users cannot delete their personal information, and they cannot limit the purposes for which the data is used. The study has designed a deep learning method that employs privacy protection technology that uses distributed collaborative learning so that multiple participants can use neural network models collaboratively without sharing the input datasets. To prevent direct leaks of personal information, participants are not shown the training datasets during the model training process, unlike traditional deep learning so that the personal information in the data can be protected. The study used a method that can selectively share subsets via an optimization algorithm that is based on modified distributed stochastic gradient descent, and the result showed that it was possible to learn with improved learning accuracy while protecting personal information.

• International journal of advanced smart convergence
• /
• v.12 no.4
• /
• pp.378-385
• /
• 2023

The importance of wearing respiratory protective equipment has been highlighted even more during the COVID-19 pandemic. Even if the suitability of respiratory protection has been confirmed through testing in a laboratory environment, there remains the potential for leakage points in the respirators due to improper application by the wearer, damage to the equipment, or sudden movements in real working conditions. In this paper, we propose a method to detect the occurrence of leak holes by measuring the pressure changes inside the mask according to the wearer's breathing activity by attaching an IoT sensor to a full-face respirator. We designed 9 experimental scenarios by adjusting the degree of leak holes of the respirator and the breathing cycle time, and acquired respiratory data for the wearer of the respirator accordingly. Additionally, we analyzed the respiratory data to identify the duration and pressure change range for each breath, utilizing this data to train a neural network model for detecting leak holes in the respirator. The experimental results applying the developed neural network model showed a sensitivity of 100%, specificity of 94.29%, and accuracy of 97.53%. We conclude that the effective detection of leak holes can be achieved by incorporating affordable, small-sized IoT sensors into respiratory protective equipment.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 1998.04a
• /
• pp.542-547
• /
• 1998

As the concern for a global energy conservation and environmental protection are increasing, it has been more important thing to correspond with CFC depletion. Alternate refrigerants have merit such as lower global warming effect, but also have demerits such as lower efficiency, miscibility, increasing noise and poor reliability problems. Then we have to develop more efficient, silent and robust compressors to satisfying world-wide demand. In this paper, parametric study on rotary compressor muffler for a room air-conditioner was carried out to investigate the effect of important design variables on noise by using Taguchi robust design method with signal-to-noise(S/N) ratio. Taguchi method seems to be helpful for finding optimum value of design variables for noise level. We also applied neural network to find optimal value of design variables.

• Journal of information and communication convergence engineering
• /
• v.2 no.4
• /
• pp.251-255
• /
• 2004

In this paper, we present a watermarking scheme based on the DWT (Discrete Wavelet Transform) and the ANN (Artificial Neural Network) to ensure the copyright protection of the digital images. The problem to embed watermark is not clear to select important coefficient in the watermarking. We used the RBF (Radial-Basis Function) to solve the problem. We didn't apply the whole wavelet coefficients, but applied to only the wavelet coefficients in the selected node. Using the ANN, although even the watermark casting process and watermark verification process are in public, nobody knows the location of embedding watermark except of authorized user. As the result, the watermark is good at the strength test-filtering, geometric transform and etc.

• ETRI Journal
• /
• v.42 no.3
• /
• pp.366-375
• /
• 2020

The network traffic prediction of a smart substation is key in strengthening its system security protection. To improve the performance of its traffic prediction, in this paper, we propose an improved gravitational search algorithm (IGSA), then introduce the IGSA into a wavelet neural network (WNN), iteratively optimize the initial connection weighting, scalability factor, and shift factor, and establish a smart substation network traffic prediction model based on the IGSA-WNN. A comparative analysis of the experimental results shows that the performance of the IGSA-WNN-based prediction model further improves the convergence velocity and prediction accuracy, and that the proposed model solves the deficiency issues of the original WNN, such as slow convergence velocity and ease of falling into a locally optimal solution; thus, it is a better smart substation network traffic prediction model.

• Proceedings of the KIEE Conference
• /
• 1995.11a
• /
• pp.131-136
• /
• 1995

This paper presents a functional ability improvement of auto-reclosing relay in the power transmission line protection. When the high speed auto-reclosing is successful, Auto-reclosing is practically valuable to improve the transient stability limit of a power system, but it is fail due to surviving fault, both electrical and mechanical stresses can result on the transformers and turbine-generator. It is true that the longer dead time of the reclosing relay gives the higher rate of successful reclosing, On the other hand, the power system does not always need high speed reclosing because of enough stability margin. This paper proposed "stability margin based dead time reclosing" in order to decrease not only the rate of unsuccessful reclosing, but the possibility of the harmful stress also. On-line transient stability assessment using artificial neural network, for implementing the proposed scheme, has studied and tested with resonable results.

• Proceedings of the KIEE Conference
• /
• 1997.07c
• /
• pp.897-899
• /
• 1997

The analysis of distribution line faults is essential to the proper protections of the power system. A high impedance fault does not make enough current to cause conventional protective devices. It is well known that undesirable operating conditions and certain types of faults on electric distribution feeders cannot be detected by using conventional protection system. This paper describes an algorithm using back-propagation neural network for pattern recognition and detection of high impedance faults. Fractal dimensions are estimated for distinction between random noise and chaotic behavior in the power system. The fractal dimension of the line current is also used as a indication of the high impedance fault.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2000.10a
• /
• pp.459-462
• /
• 2000

The analysis of distribution line faults is essential to the proper protection of power system. A high impedance fault(HIF) dose not make enough current to cause conventional protective device operating. so it is well hon that undesirable operating conditions and certain types of faults on electric distribution feeders cannot be detected by using conventional protection system. In this paper, we prove that the nature of the high impedance faults is indeed a deterministic chaos, not a random motion Algorithms for estimating Lyapunov spectrum and the largest Lyapunov exponent are applied to various fault currents detections in order to evaluate the orbital instability peculiar to deterministic chaos dynamically, and fractal dimensions of fault currents which represent geometrical self-similarity are calculated. Wavelet transform analysis is applied the time-scale information to fault signal. Time-scale representation of high impedance faults can detect easily and localize correctly the fault waveform.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.9 no.5
• /
• pp.538-544
• /
• 1999

In this paper, we propose a new fault detection algorithm for transmission line protection using ANFIS(Adaptive Network Fuzzy Inference System). The developed system consists of two subsystems: fault type classification, and fault location estimation. We use rms value, zero sequence component and positive sequence of current, and then using learning method of neural network, premise and consequent parameters are tuned properly. To prove the performance of the proposcd system, generated data by EMTP(Electr0- Magnetic Transient Program) sin~ulationi s used. It is shown that the proposed relaying classifies fault types accurately and advances fault location estimation.

• Journal of Korean Neurosurgical Society
• /
• v.29 no.2
• /
• pp.167-179
• /
• 2000

Objective : We studied to clarify the effective time zone of mild hypothermic neural protection during ischemia and/or reperfusion after middle cerebral artery occlusion. Methods : In a reversible cerebral infarct model which maintained reperfusion of blood flow after middle cerebral artery occlusion for two hours, the size of cerebral infarction, cerebral edema and the extent of neurological deficit were observed and analyzed for comparison between the control and the experimental groups under hypothermia($33.5^{\circ}C$). The temporalis muscle temperature was reduced to $33.5^{\circ}C$ by surface cooling for two hours during middle cerebral artery occlusion for study group I. The following groups applied hypothermia for two-hour periods after reperfusion : group II(0-2 hours), group III(2-4 hours), and group IV(4-6 hours). They were rewarmed to $36.5^{\circ}C$ until sacrified at 2, 4, 6, 12, and 24 hours after reperfusion. Control group was maintained at normothermia without hypothermia. Results : In the experimental groups with hypothermia, the average value of the size of cerebral infarction($mean{\pm}SD$) was $1.97{\pm}1.65%$, which was a remarkable reduction over that of the control, $4.93{\pm}3.79%$. In the control, a progressive increase was shown in the size of infarction from point of reperfusion to 6 hours after reperfusion without further changes in size afterward. Intra-ischemic hypothermia(group I) prevented ischemic injury but did not prevent reperfusion injury. Group II examplified the most neural protective effect in comparison to the control group and group IV(p<0.05). The cortex was more vulnerable to reperfusion injury than the subcortex. Mild hypothermia showed more neural protective effects on the cortex than subcortex. Conclusion : The most appropriate time zone for application of mild hypothermia was defined to be within four hours following reperfusion.