• Communications for Statistical Applications and Methods
• /
• v.30 no.5
• /
• pp.501-516
• /
• 2023

Knowledge distillation (KD) refers to extracting knowledge from a large and complex model (teacher) and transferring it to a relatively small model (student). This can be done by training the teacher model to obtain the activation function values of the hidden or the output layers and then retraining the student model using the same training data with the obtained values. Recently, relational KD (RKD) has been proposed to extract knowledge about relative differences in training data. This method improved the performance of the student model compared to conventional KDs. In this paper, we propose a new method for RKD by introducing a new loss function for RKD. The proposed loss function is defined using the area difference between the teacher model and the student model in a specific hidden layer, and it is shown that the model can be successfully compressed, and the generalization performance of the model can be improved. We demonstrate that the accuracy of the model applying the method proposed in the study of model compression of audio data is up to 1.8% higher than that of the existing method. For the study of model generalization, we demonstrate that the model has up to 0.5% better performance in accuracy when introducing the RKD method to self-KD using image data.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.37 no.8
• /
• pp.1060-1074
• /
• 2013

This study measured and analyzed pressure at each measurement part of imported compression stockings sold in Korea to provide basic information to establish a pressure standard and grade ranking. This study used 40 medical compression stockings imported from 6 countries. Pressure measurements were taken at 11 points: front side and back side of ankle, end-point of the gastrocnemius muscle, front, inner side, back, and outer side of calf, back side of below knew girth, inner side, and outer side of mid-thigh girth, and inner side of thigh girth. AMI 3037-10 and AMI 3037-2 were used for measurements taken inside an environmental chamber at a temperature of $21^{\circ}C$ and a relative humidity (RH) of 65%. For the measurements, 11 air pack sensors were attached to a wooden model leg (Hohenstein) and three measurements were taken at each measurement point in three minutes. The average of these measurements was used for analysis. The findings of this study were as follows. As for the front side of the ankle, of the 40 products, 14 products (6 USA, 2 Swiss, 3 Italian, and 2 Taiwanese) were within the pressure range indicated on the product label; however, no German products fell within the pressure range. A total of 8 products (5 USA, 1 Swiss, 1 Italian, and 1 German) were gradient compression type; however, no Japanese or Taiwanese product were of this type. The majority of products had the highest pressure at the end-point of the gastrocnemius muscle. Only 3 products, 1 USA (Jobst Opaque 30-40mmHg), 1 Swiss (Sigvaris Cotton 34-46mmHg) and 1 Italian (Jobstocking 25-32mmHg), had measurements that met the indicated standard pressure, were a gradient compression type, and met the overall standard for compression stockings.

• Computers and Concrete
• /
• v.11 no.2
• /
• pp.123-148
• /
• 2013

The nonlinear finite element method with eight noded isoparametric quadrilateral element for concrete and two noded element for reinforcement is used for the prediction of the behavior of reinforcement concrete structures. The disturbed state concept (DSC) including the hierarchical single surface (HISS) plasticity model with associated flow rule with modifications is used to characterize the constitutive behavior of concrete both in compression and in tension which is named DSC/HISS-CT. The HISS model is applied to shows the plastic behavior of concrete, and DSC for microcracking, fracture and softening simulations of concrete. It should be noted that the DSC expresses the behavior of a material element as a mixture of two interacting components and can include both softening and stiffening, while the classical damage approach assumes that cracks (damage) induced in a material treated acts as a void, with no strength. The DSC/HISS-CT is a unified model with different mechanism, which expresses the observed behavior in terms of interacting behavior of components; thus the mechanism in the DSC is much different than that of the damage model, which is based on physical cracks which has no strength and interaction with the undamaged part. This is the first time the DSC/HISS-CT model, with the capacity to account for both compression and tension yields, is applied for concrete materials. The DSC model allows also for the characterization of non-associative behavior through the use of disturbance. Elastic perfectly plastic behavior is assumed for modeling of steel reinforcement. The DSC model is validated at two levels: (1) specimen and (2) practical boundary value problem. For the specimen level, the predictions are obtained by the integration of the incremental constitutive relations. The FE procedure with DSC/HISS-CT model is used to obtain predictions for practical boundary value problems. Based on the comparisons between DSC/HISS-CT predictions, test data and ANSYS software predictions, it is found that the model provides highly satisfactory predictions. The model allows computation of microcracking during deformation leading to the fracture and failure; in the model, the critical disturbance, Dc, identifies fracture and failure.

• Journal of the Korean Wood Science and Technology
• /
• v.34 no.5
• /
• pp.19-28
• /
• 2006

There has been considerable research in recent times in light-timber med structures in fires. These structures have included horizontal (floor-like) panels in bending and walls under eccentric and approximately concentric vertical loading. It has been shown that compression properties are the most dominant mechanical properties in affecting structural response of these structures in fire. Compression properties have been obtained by various means as functions of one variable only, temperature. It has always been expected that compression properties would be significantly affected by moisture and stress, as well. However, these variables have been largely ignored to simplify the complex problem of predicting the response of light-timber framed structures in fire. Full-scale experiments on both the panels and walls have demonstrated the high level of significance of moisture and stress for a limited range of conditions. Described in this paper is an overview of these conditions and experiments undertaken to obtain compression properties as a functions of moisture, stress and temperature. The experiments limited temperatures to $20{\sim}100^{\circ}C$. At higher temperatures moisture vaporizes and moisture and stress are less significant. Described also is a creep model for wood at high temperatures.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.21 no.11
• /
• pp.1403-1412
• /
• 1997

When a high-speed railway train enters a tunnel, a compression wave is generated ahead of the train and propagates through the tunnel, compressing and accelerating the rest air in front of the wave. At the exit of the tunnel, an impulsive wave is emitted outward toward the surrounding, which causes a positive impulsive noise like a kind of sonic boom produced by a supersonic aircraft. With the advent of high-speed train, such an impulsive noise can be large enough to cause the noise problem, unless some attempts are made to alleviate its pressure levels. In the purpose of the impulsive noise reduction, the present study calculated the effect of porous walls on the compression wave propagating into a model tunnel. Two-dimensional unsteady compressible equations were differenced by using a Piecewise Linear Method. Calculation results show that the cavity/porous wall system is very effective for a compression wave with a large nonlinear effect. The porosity of 30% is most effective for the reduction of the maximum pressure gradient of the compression wave front. The present calculation results are in a good agreement with experimental ones obtained previously.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.20 no.12
• /
• pp.4036-4043
• /
• 1996

When a high-speed railway train enters a tunnel, a compression wave is generated ahead of the train and propagates along the tunnel, compressing and accelerating the rest air in front of the wave. At the exit of the tunnel, an impulsive wave is emitted outward toward the surrounding, which causes a positive impulsive noise like a kind of sonic boom produced by a supersonic aircraft. With the advent of high-speed train, such an impulsive noise can be large enough to cause the noise problem, unless some attempts are made to alleviate its pressure levels. In the purpose of the impulsive noise reduction, the present study tested the effect of porous walls on the compression wave propagating into a model tunnel. Experimental results were obtained using a shock tube with an open end. The results showed that the cavity/porous wall is very effective for the compression wave with a large nonlinear effect. The porosity of 30% is most effective for attenuation and pressure gradient reduction of the compression wave front. Also the impulsive noise reduction increases with increasing the length and height of the cavity, compared with the tunnel equivalent diameter.

• Steel and Composite Structures
• /
• v.46 no.2
• /
• pp.153-173
• /
• 2023

Rectangular concrete-filled steel tubular (RCFST) column, a type of concrete-filled steel tubular (CFST), is widely used in compression members of structures because of its advantages. This paper proposes a robust machine learning-based framework for predicting the ultimate compressive strength of RCFST columns under both concentric and eccentric loading. The gradient boosting neural network (GBNN), an efficient and up-to-date ML algorithm, is utilized for developing a predictive model in the proposed framework. A total of 890 experimental data of RCFST columns, which is categorized into two datasets of concentric and eccentric compression, is carefully collected to serve as training and testing purposes. The accuracy of the proposed model is demonstrated by comparing its performance with seven state-of-the-art machine learning methods including decision tree (DT), random forest (RF), support vector machines (SVM), deep learning (DL), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), and categorical gradient boosting (CatBoost). Four available design codes, including the European (EC4), American concrete institute (ACI), American institute of steel construction (AISC), and Australian/New Zealand (AS/NZS) are refereed in another comparison. The results demonstrate that the proposed GBNN method is a robust and powerful approach to obtain the ultimate strength of RCFST columns.

• Structural Engineering and Mechanics
• /
• v.84 no.2
• /
• pp.239-251
• /
• 2022

The use of lateral reinforcement in confined concrete columns can improve bearing capacity and deformability. The lateral responses of lateral reinforcement significantly influence the effective confining pressure on core concrete. However, lateral strain-axial strain model of concrete columns confined by lateral reinforcement has not received enough attention. In this paper, based on experimental results of 85 concrete columns confined by lateral reinforcement under axial compression, the effect of unconfined concrete compressive strength, volumetric ratio, lateral reinforcement yield strength, and confinement type on lateral strain-axial strain curves was investigated. Through parameter analysis, it indicated that with the same level of axial strain, the lateral strain slightly increased with the increase in the unconfined concrete compressive strength, but decreased with the increase in volumetric ratio significantly. The lateral reinforcement yield strength had slight influence on lateral strain-axial strain curves. At the same level of lateral strain, the axial strain of specimen with spiral was larger than that of specimen with stirrup. Furthermore, a lateral strain-axial strain model for concrete columns confined by lateral reinforcement under axial compression was proposed by introducing the effects of unconfined concrete compressive strength, volumetric ratio, confinement type and effective confining pressure, which showed good agreement with the experimental results.

• Proceedings of the Korea Concrete Institute Conference
• /
• 2008.11a
• /
• pp.81-84
• /
• 2008

The purposes of this study are to verify a reasonable model of material characteristic and to propose a rational model of reinforcement characteristic considering monotonic and cyclic loading about manufactured reinforcing steel in Korea. Longitudinal reinforcements of the plastic hinge region were behaved tensile deformation and compressional deformation by direction of lateral loading. However Confinement steels were behaved only tensile deformation by lateral loading. Transverse steels were laid the state of tension in the lateral loading of time, and they were laid state that stress is zero when it was removed lateral load. The tests for cyclic tension loading were performed for test variable as yield strength and reinforcement bar sizes. It was estimated that the total strain energy per unit volume was 74 $MJ/m^3$. The modified ultimate concrete compression strain model was proposed based on experimental study of cyclic tension test for manufactured reinforcing steel in Korea.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.26 no.7
• /
• pp.1324-1331
• /
• 2002

Densification behavior of composite powders was investigated during cold compaction. Experimental data were obtained for aluminum alloy powder mixed with zirconia powder inclusion under triaxial compression. The Cap model with constraint factors was implemented into a finite element program (ABAQUS) to simulate compaction responses of composite powders during cold compaction. Finite element results were compared with experimental data for densification behavior of composite powders under cold isostatic pressing and die compaction. The agreements between experimental data and finite element calculations from the Cap model with constraint factors were good.