• Journal of the Ergonomics Society of Korea
• /
• v.15 no.1
• /
• pp.91-103
• /
• 1996

For determining thermal comfort properties of work suit in an atomic power plant, three different coverall ensembles (PVE, PET/Rayon, PP Nonwoven) were selected and the resistance to dry and evaporative heat transfer were measured for each ensemble by using a sweating thermal manikin. Also, PMV (Predicted Mean Vote) and PPD(Predicted Percentage of Dissatisfied) indices were predicted according to ISO 7730. As a result, ideal environmental conditions in an atomic power plant were suggested to make workers feel thermally comfortable. In addition, ideal intrinsic insulation values of coverall ensembles as a work suit under the present environmental conditions in the at6omic power plant were provided. The information given in this paper can be used to control environmental conditions in the atomic power plant thermally comfortable and to select a proper work suit for providing thermal comfort to the workers.

• Proceedings of the IEEK Conference
• /
• 2002.06a
• /
• pp.323-326
• /
• 2002

In this paper, we study the improved bounds on the performance of low-density parity-check (LDPC) codes over binary-input additive white Gaussian noise (AWGN) channels with belief propagation (BP) decoding in log domain. We define an extended Gallager ensemble based on a new method of constructing parity check matrix and make use of this way to improve upper bound of LDPC codes. At the same time, many simulation results are presented in this paper. These results indicate the extended Gallager ensembles based on Hamming codes have typical minimum distance ratio, which is very close to the asymptotic Gilbert Varshamov bound and the superior performance which is better than the original Gallager ensembles.

• Communications for Statistical Applications and Methods
• /
• v.13 no.2
• /
• pp.449-466
• /
• 2006

We propose to use variable selection methods based on penalized regression for pruning decision tree ensembles. Pruning methods based on LASSO and SCAD are compared with the cluster pruning method. Comparative studies are performed on some artificial datasets and real datasets. According to the results of comparative studies, the proposed methods based on penalized regression reduce the size of boosting ensembles without decreasing accuracy significantly and have better performance than the cluster pruning method. In terms of classification noise, the proposed pruning methods can mitigate the weakness of AdaBoost to some degree.

• Journal of the Korean Ceramic Society
• /
• v.40 no.11
• /
• pp.1027-1046
• /
• 2003

Nanostructured materials can be engineered by the controlled assembly of several suitable nano-objects as the building blocks. While, materials properties are determined by their atomic and molecular constituents and structure, their functionalities emerge when the microstructure of these early ensembles is in the nanometer regime. The properties and functionalities of these ensembles may be different as their size grows from the nano-regime to the micron regime and bulk structures. Nanotechnology, offers a unique possibility to manipulate the properties through the fabrication of materials using the nano-objects as building blocks. Nanotechnology is therefore considered an enabling technology by which existing materials, virtually all man-made materials, can acquire novel properties and functionalities making them suitable for numerous novel applications varying from structural and functional to advanced biomedical in-vivo and in-vitro applications.

• Information Systems Review
• /
• v.24 no.1
• /
• pp.39-57
• /
• 2022

This paper proposes the GMOPTBoost algorithm to improve the performance of the AdaBoost algorithm for bankruptcy prediction in which class imbalance problem is inherent. AdaBoost algorithm has the advantage of providing a robust learning opportunity for misclassified samples. However, there is a limitation in addressing class imbalance problem because the concept of arithmetic mean accuracy is embedded in AdaBoost algorithm. GMOPTBoost can optimize the geometric mean accuracy and effectively solve the category imbalance problem by applying Gaussian gradient descent. The samples are constructed according to the following two phases. First, five class imbalance datasets are constructed to verify the effect of the class imbalance problem on the performance of the prediction model and the performance improvement effect of GMOPTBoost. Second, class balanced data are constituted through data sampling techniques to verify the performance improvement effect of GMOPTBoost. The main results of 30 times of cross-validation analyzes are as follows. First, the class imbalance problem degrades the performance of ensembles. Second, GMOPTBoost contributes to performance improvements of AdaBoost ensembles trained on imbalanced datasets. Third, Data sampling techniques have a positive impact on performance improvement. Finally, GMOPTBoost contributes to significant performance improvement of AdaBoost ensembles trained on balanced datasets.

• Journal of Intelligence and Information Systems
• /
• v.5 no.1
• /
• pp.95-101
• /
• 1999

In this paper, a comparison between different methods to combine predictions from neural networks will be given. These methods are bagging, bumping, and balancing. Those are based on the analysis of the ensemble generalization error into an ambiguity term and a term incorporating generalization performances of individual networks. Neural Networks and AI machine learning models are prone to overfitting. A strategy to prevent a neural network from overfitting, is to stop training in early stage of the learning process. The complete data set is spilt up into a training set and a validation set. Training is stopped when the error on the validation set starts increasing. The stability of the networks is highly dependent on the division in training and validation set, and also on the random initial weights and the chosen minimization procedure. This causes early stopped networks to be rather unstable: a small change in the data or different initial conditions can produce large changes in the prediction. Therefore, it is advisable to apply the same procedure several times starting from different initial weights. This technique is often referred to as training ensembles of neural networks. In this paper, we presented a comparison of three statistical methods to prevent overfitting of neural network.

• Fashion & Textile Research Journal
• /
• v.11 no.6
• /
• pp.947-952
• /
• 2009

The purpose of this study is to analyze skin wettedness($w$) used as the rate index of thermal comfort, and to evaluate the wear comfort of outdoorwear. Skin wettedness is widely used to express the degree of thermal comfort. If skin wettedness exceeds a certain threshold, the body feels damp and discomfort. An experiment which consisted of rest(30 min), exercise(30 min) and recovery(20 min) periods was administered in a climate chamber with 10 healthy male participants. Two kinds of outdoorwears made of 100% cotton fabrics (Control) and specially engineered fabrics having feature of quick sweat absorbency and high speed drying fabric (Functional) were evaluated in the experiment. The condition of climate chamber was controlled according to the thermal insulation of 4 kinds of experimental ensembles(E1~E4). Total sweat loss, sweat loss absorbed into clothing and skin temperature were measured. Skin wettedness was calculated from the ratio of evaporative rate to the maximal evaporative capacity. Skin wettedness of 'Functional' was lower than 'Control' in the 3 kinds of ensembles(E1, E2, E4) because the materials of 'Functional' were composed of quick sweat absorbency and high speed drying fabrics, water vapour permeability and waterproof fabrics.

• Bulletin of the Korean Chemical Society
• /
• v.30 no.2
• /
• pp.454-458
• /
• 2009

Polypyrrole (PPy) nanotubes have been synthesized by chemical oxidative polymerization of pyrrole within the pores of polycarbonate membrane using the technology of diffusion of solutes. The nanotubes array prepared by the proposed method can be considered as nanoelectrode ensembles (NEEs). An amperometric uric acid sensor based on PPy NEEs has been developed and used for determination of uric acid in human serum samples. The electrode can direct response to uric acid at potential of 0.60V vs. SCE with wide linear range of $1.52{\times}10^{-6}\;to\;1.54{\times}10^{-3}\;M.\;The\;detection\;limit\;is \;3.02{\times}10^{-7}$ M. This sensor has been used to determine uric acid in real serum samples. PPy NEEs is thought of as a good application in the foreground.

• Smart Structures and Systems
• /
• v.26 no.3
• /
• pp.263-275
• /
• 2020

This research is dedicated to slope stability analysis using novel intelligent models. By coupling a neural network with spotted hyena optimizer (SHO), salp swarm algorithm (SSA), shuffled frog leaping algorithm (SFLA), and league champion optimization algorithm (LCA) metaheuristic algorithms, four predictive ensembles are built for predicting the factor of safety (FOS) of a single-layer cohesive soil slope. The data used to develop the ensembles are provided from a vast finite element analysis. After creating the proposed models, it was observed that the best population size for the SHO, SSA, SFLA, and LCA is 300, 400, 400, and 200, respectively. Evaluation of the results showed that the combination of metaheuristic and neural approaches offers capable tools for estimating the FOS. However, the SSA (error = 0.3532 and correlation = 0.9937), emerged as the most reliable optimizer, followed by LCA (error = 0.5430 and correlation = 0.9843), SFLA (error = 0.8176 and correlation = 0.9645), and SHO (error = 2.0887 and correlation = 0.8614). Due to the high accuracy of the SSA in properly adjusting the computational parameters of the neural network, the corresponding FOS predictive formula is presented to be used as a fast yet accurate substitution for traditional methods.

• Nuclear Engineering and Technology
• /
• v.55 no.10
• /
• pp.3700-3708
• /
• 2023

The stochastic origin ensembles method with resolution recovery (SOE-RR) has been proposed to reconstruct proton-induced prompt gammas (PGs), and the reconstructed PG image was used for range verification. However, due to low detection efficiency, the number of valid events is low. Such a low-count condition can degrade the accuracy of the SOE-RR method for proton range verification. In this study, we proposed two strategies to improve the reconstruction of the SOE-RR algorithm for low-count PG imaging. We also studied the number of iterations and repetitions required to achieve reliable range verification. We simulated a proton beam (10⁸ protons) irradiated on a water phantom and used a two-layer Compton camera to detect 4.44-MeV PGs. Our simulated results show that combining the SOE-RR algorithm with restricted volume (SOE-RR-RV) can reduce the error of the estimation of the Bragg peak position from 5.0 mm to 2.5 mm. We also found that the SOE-RR-RV algorithm initialized using a back-projection image could improve the convergence rate while maintaining accurate range verification. Finally, we observed that the improved SOE-RR algorithm set for 60,000 iterations and 25 repetitions could provide reliable PG images. Based on the proposed reconstruction strategies, the SOE-RR algorithm has the potential to achieve a positioning error of 2.5 mm for low-count PG imaging.