• Smart Media Journal
• /
• v.12 no.4
• /
• pp.19-26
• /
• 2023

IoT edge services utilizing neuromorphic hardware architectures are suitable for autonomous IoT applications as they perform intelligent processing on the device itself. However, spiking neural networks applied to neuromorphic hardware are difficult for IoT developers to comprehend due to their complex structures and various hyper-parameters. In this paper, we propose a method for generating spiking neural network (SNN) models that satisfy user performance requirements while considering the constraints of neuromorphic hardware. Our proposed method utilizes previously trained models from pre-processed data to find optimal SNN model parameters from profiling data. Comparing our method to a naive search method, both methods satisfy user requirements, but our proposed method shows better performance in terms of runtime. Additionally, even if the constraints of new hardware are not clearly known, the proposed method can provide high scalability by utilizing the profiled data of the hardware.

• Korean Journal of Environmental Agriculture
• /
• v.14 no.3
• /
• pp.302-311
• /
• 1995

Benfuresate or oxolinic acid, as an experimental pesticide, was applied to the different textural paddy or upland soil respectively under the field condition and the residual concentrations were determined. Six kinetic models were employed to characterize the best-fit kinetic model describing the residual pattern of benfuresate or oxolinic acid and the $t\frac{1}{2}$ estimated from each model was comparatively assessed. All of the six models explained significantly the residual patterns of the pesticides but the empirical models such as PF, EL, and PB were not recommendable for the $t\frac{1}{2}$ estimation. Among theoretical models, the residual patterns were followed in the orders of the second-order(SO)>first-order(FO)>zero-order(ZO) kinetics, judging from the size and significance of coefficient of determination and standard error. However, the multiple FO model, consisting of the fast and slow decomposition steps, was better than the single FO model for the residual pattern and the $r^2$ in this case became similar to that of SO kinetic model. Thus the multiple FO and SO models were represented as the best fit model of the experimental pesticide. The $t\frac{1}{2}$ of benfuresate estimated from the single FO kinetic model in Weolgog and Cheongwon series was 49 and 63 days, respectively, which were 20 and 13% longer than the respective $t\frac{1}{2}$ from the SO kinetic model. The $t\frac{1}{2}$ of oxolinic acid from the FO model in Yonggye and Ihyeon series were 87 and 51% longer than those from the SO kinetic model, respectively. These results demonstrated that the best-fit model representing the residual pattern of a pesticide and the resultant $t\frac{1}{2}$ might be variable with the kinds of pesticides and the environmental conditions. Therefore it is recommended that the half-life of a pesticide be assessed from the best-fit model rather than from the FO kinetic model uniformly.

• Special Issue of the Society of Naval Architects of Korea
• /
• 2015.09a
• /
• pp.50-55
• /
• 2015

Application of newly conceptualized Anti-Splash Devices designed for COT vent pipes were studied on a P/V valve located on the upper deck of an oil carrier vessel. Anti-Splash devices are used in the shipbuilding industry in order to avoid oil overflow and spray accidents caused by excess pressure and vacuum condition in the cargo oil tanks. These conditions are caused by the transverse and longitudinal sloshing forces that arise from ship motion during sea voyages. The main issue with existing Anti-Splash device model is flux at the outlet of the Anti-Splash Device, and so, new conceptual models for the Anti-Splash device were developed and compared to existing Anti-Splash device model using CFD analysis. Transient analysis was used to capture the flow and velocity of each model and a comparative analysis was performed between old and new-concept models. This data was used to determine the optimal design parameters in order to develop an optimized Anti-Splash Device. A Factory acceptance test was performed on the new-concept models in order to verify the performance and efficiency against their design requirements and other criterion. The final step performed was to apply the optimized Anti-Splash Device models for COT vent pipes to an actual vessel and verify performance through a seawater cargo operation during a sea voyage as per the ship owner's request. The patent for the aforementioned device was obtained by the Korean Intellectual property Office dated Dec. 18th,2014.

• Geomechanics and Engineering
• /
• v.6 no.1
• /
• pp.1-15
• /
• 2014

In this study, artificial neural network (ANN) and multiple regression (MR) models were developed to predict the critical factor of safety ($F_s$) of the homogeneous finite slopes subjected to earthquake forces. To achieve this, the values of $F_s$ in 5184 nos. of homogeneous finite slopes having different slope, soil and earthquake parameters were calculated by using the Simplified Bishop method and the minimum (critical) $F_s$ for each of the case was determined and used in the development of the ANN and MR models. The results obtained from both the models were compared with those obtained from the calculations. It is found that the ANN model exhibits more reliable predictions than the MR model. Moreover, several performance indices such as the determination coefficient, variance account for, mean absolute error, root mean square error, and the scaled percent error were computed. Also, the receiver operating curves were drawn, and the areas under the curves (AUC) were calculated to assess the prediction capacity of the ANN and MR models developed. The performance level attained in the ANN model shows that the ANN model developed can be used for predicting the critical $F_s$ of the homogeneous finite slopes subjected to earthquake forces.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.15 no.6
• /
• pp.754-760
• /
• 2012

To improve the efficiency and economics of national defence, the needs for defence M&S systems are greatly increased, but the lack of interoperability and reusability lowers the level of reliability in the existing simulations models. Recently, many research activities have been performed and some conceptual models, such as CMMS, DCMF, and KAMA, were introduced. However, these models don't support the whole process of defence M&S system development nor provide integrated domain knowledge and up-to-date data. In this paper, we analyzed the analysts' guidances and doctrines, which are the basic ingredients for the simulation system development, and proposed a systematic structure for all the defence M&S stakeholders to share the standardized domain knowledge, modeling algorithm and data. This will be a logical bridge between Mission Space and Simulation Space, and provide substantial data sets to build simulation systems.

• Korean Journal of Materials Research
• /
• v.33 no.7
• /
• pp.273-278
• /
• 2023

In this study, based on the saturation magnetic flux density experimental values (B_s) of 622 Fe-based bulk metallic glasses (BMGs), regression models were applied to predict B_s using artificial neural networks (ANN), and prediction performance was evaluated. Model performance evaluation was investigated by using the F1 score together with the coefficient of determination (R² score), which is mainly used in regression models. The coefficient of determination can be used as a performance indicator, since it shows the predicted results of the saturation magnetic flux density of full material datasets in a balanced way. However, the BMG alloy contains iron and requires a high saturation magnetic flux density to have excellent applicability as a soft magnetic material, and in this study F1 score was used as a performance indicator to better predict B_s above the threshold value of B_s (1.4 T). After obtaining two ANN models optimized for the R² and F1 score conditions, respectively, their prediction performance was compared for the test data. As a case study to evaluate the prediction performance, new Fe-based BMG datasets that were not included in the training and test datasets were predicted using the two ANN models. The results showed that the model with an excellent F1 score achieved a more accurate prediction for a material with a high saturation magnetic flux density.

• The Journal of the Korea Contents Association
• /
• v.12 no.10
• /
• pp.357-364
• /
• 2012

Dental scanner-based dental digital models have been developed that have the potential to replace conventional stone model. The aim of this study was examine the accuracy and reliability of measurements made on digital models. A master model with the prepared upper full arch tooth was used. Stone model(N=10) were produced from master model, and on the other hands, digital models were made with the white light scanner(Identica, korea). One examiner individually measured 6 parameters on the conventional model and the digital models on two occasions. The student's t-test for paired samples and intra-class correlation coefficient(ICC) were used for statistical analysis. At the intra-examiner reliability of measurement, ICC at the stone and digital models were 0.75 and 0.87. The mean difference between measurements made directly on the stone models and those made on the digital models was 0.11-0.23mm, and was statistically significant(P<0.05). These in vitro studies show that accuracy of the digital model is similar to that of the stone model. These results will have to be confirmed in further clinical studies.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2005.09a
• /
• pp.257-258
• /
• 2005

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2007.06a
• /
• pp.123-124
• /
• 2007

• Journal of the Korean Institute of Telematics and Electronics A
• /
• v.31A no.1
• /
• pp.54-60
• /
• 1994

In this paper, we present the analytical models for the change of the lateral electric field distribution and the velocity saturation region length with the electron trapping of stressed SC-PMOSFET in the saturation region. To derive the hot-electron-induced lateral electric field of stressed SC-PMOSFET. Ko's pseudo two dimensional box model in the saturation region which illustrates the analysis of the velocity saturation region is modified under the condition of electron trapping in the oxide near the drain region. From the results, we have the following lateral electric field in the y-direction, that is, E(y) ES1satT.cosh(y/l) qNS1tT.sinh(y/l)/lCox. It is shown that the trapped electrons influence the field in the drain region. decreasing the lateral electric field. Calculated velocity saturaion length increases with the trapped electrons. increasing the drain current of stressed SCPMOSFET. This results well explain the HEIP phenomenon of PMOSFET's.