• The Journal of the Korea Contents Association
• /
• v.18 no.8
• /
• pp.525-535
• /
• 2018

If the frequency of a particular class is excessively higher than the frequency of other classes in the classification problem, data imbalance problems occur, which make machine learning distorted. Corporate bankruptcy prediction often suffers from data imbalance problems since the ratio of insolvent companies is generally very low, whereas the ratio of solvent companies is very high. To mitigate these problems, it is required to apply a proper sampling technique. Until now, oversampling techniques which adjust the class distribution of a data set by sampling minor class with replacement have popularly been used. However, they are a risk of overfitting. Under this background, this study proposes ROSE(Random Over Sampling Examples) technique which is proposed by Menardi and Torelli in 2014 for the effective corporate bankruptcy prediction. The ROSE technique creates new learning samples by synthesizing the samples for learning, so it leads to better prediction accuracy of the classifiers while avoiding the risk of overfitting. Specifically, our study proposes to combine the ROSE method with SVM(support vector machine), which is known as the best binary classifier. We applied the proposed method to a real-world bankruptcy prediction case of a Korean major bank, and compared its performance with other sampling techniques. Experimental results showed that ROSE contributed to the improvement of the prediction accuracy of SVM in bankruptcy prediction compared to other techniques, with statistical significance. These results shed a light on the fact that ROSE can be a good alternative for resolving data imbalance problems of the prediction problems in social science area other than bankruptcy prediction.

• Journal of radiological science and technology
• /
• v.47 no.1
• /
• pp.29-37
• /
• 2024

This study aims to evaluate the performance of the U-Net based learning model that may vary depending on the histogram equalization algorithm. The subject of the experiment were 17 radiology students of this college, and 1,727 data sets in which the region of interest was set in the thyroid after acquiring ultrasound image data were used. The training set consisted of 1,383 images, the validation set consisted of 172 and the test data set consisted of 172. The equalization algorithm was divided into Histogram Equalization(HE) and Contrast Limited Adaptive Histogram Equalization(CLAHE), and according to the clip limit, it was divided into CLAHE8-1, CLAHE8-2. CLAHE8-3. Deep Learning was learned through size control, histogram equalization, Z-score normalization, and data augmentation. As a result of the experiment, the Attention U-Net showed the highest performance from CLAHE8-2 to 0.8355, and the U-Net and BSU-Net showed the highest performance from CLAHE8-3 to 0.8303 and 0.8277. In the case of mIoU, the Attention U-Net was 0.7175 in CLAHE8-2, the U-Net was 0.7098 and the BSU-Net was 0.7060 in CLAHE8-3. This study attempted to confirm the effects of U-Net, Attention U-Net, and BSU-Net models when histogram equalization is performed on ultrasound images. The increase in Clip Limit can be expected to increase the ROI match with the prediction mask by clarifying the boundaries, which affects the improvement of the contrast of the thyroid area in deep learning model learning, and consequently affects the performance improvement.

• International Journal of Automotive Technology
• /
• v.4 no.2
• /
• pp.87-94
• /
• 2003

Recent advancements of permanent magnet (PM) materials and solid-state devices have contributed to a substantial performance improvement of permanent magnet machines. Owing to the rare-earth PMs, these motors have higher efficiency, power factor, output power per mass and volume, and better dynamic performance than induction motors without sacrificing reliability. Not surprisingly, they are continuously receiving serious considerations for a variety of automotive and propulsion applications. An electric vehicle (EV) requires a high-effficient propulsion system having a wide operating range and a capability of generating a high peak torque for short durations. The improvement of torque-speed performance for these systems is consequently very important, and researches in various aspects are therefore being actively pursued. A great emphasis has been placed on the efficiency and optimal utilization of PM machines. This requires attention to many aspects related to the machine design and overall performance. In this respect, the prediction of iron losses is particularly indispensable and challenging, especially for drives with a deep field-weakening range. The objective of this paper is to present iron loss estimations of a PM motor over a wide speed range. As aforementioned, in EV applications core losses can be significant during high-speed operation and it is imperative to evaluate these losses accurately and take them into consideration during the motor design stage. In this investigation, the losses are predicted by using an analytical model and a 2D time-stepped finite element method (FEM). The results from different analytical approaches are compared with the FEM computations. The validity of each model is then evaluated by these comparisons.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.2C
• /
• pp.110-120
• /
• 2003

In this paper, we propose a Multiresolution Motion Estimation(MRME) adapted to Low-Band-Shift(LBS) method in wavelet domain. To overcome shift-variant property on wavelet coefficients, the LBS was previously proposed. This method which is applied to reference frame in video coding technique, has superior performance in terms of rate-distortion characteristic. However, this method needs more memory and computational complexity. In this paper, The computational complexity of the proposed method(LBS-MRME) is about 15.6% of that of existing method at 3-level wavelet transform. And although it has about 7 times as much as existing method's motion vector since each subband has different motion vector, it decreases motion compensated prediction error by detailed motion estimation, and then has better efficient coding performance. The experimental results with the proposed method showed about 0.3∼11.6% improvement of MAD performance in case of lossless coding, and 0.3∼3.0㏈ improvement of PSNR performance at the same bit rate in case of lossy coding.

• Journal of Korea Game Society
• /
• v.11 no.1
• /
• pp.101-109
• /
• 2011

The performance of server becomes important issues in online game with the online game market expansion. This paper proposes a method of improving performance to decrease synchronized packets for each entity's informations in game. Our method provides adapted solution of reconstructing spatial subdivision to reduce a load of movement between boundary regions using prediction of entity's movement range and disabled regions where entity can not move to. It is shown through the experiments that proposed method outperforms existing method in terms of processing quantity of packets.

• Journal of Broadcast Engineering
• /
• v.26 no.2
• /
• pp.208-220
• /
• 2021

This paper evaluates the performance of multilayer VVC and SHVC. Multilayer VVC supports a multi-layer coding and many coding technologies have been added and extended compared to SHVC. For this reason, it is necessary to evaluate the multi-layer coding performance of VVC and the coding performance for inter-layer reference prediction. Multilayer VVC provides significant BD-rate improvement of AI 24.4%, RA 29.4%, LDB 29.4%, LDP 32.6% on average when compared to SHVC, so that it is shown that VVC can provide scalability more efficiently. On the other hand, the complexity of the encoding time increases by an average of 14 times and decoding time by an average of 1.8 times, which requires efforts to reduce the complexity.

• Earthquakes and Structures
• /
• v.11 no.5
• /
• pp.779-807
• /
• 2016

Ground Motion Prediction Equations (GMPEs) are essential tools in seismic hazard analysis. With the introduction of probabilistic approaches for the estimation of seismic response of structures, also known as, performance based earthquake engineering framework; new tasks are defined for response spectrum such as the reference criterion for effective structure-specific selection of ground motions for nonlinear time history analysis. One of the recent efforts to introduce a high quality databank of ground motions besides the corresponding selection scheme based on the broadband spectral consistency is the development of SIMBAD (Selected Input Motions for displacement-Based Assessment and Design), which is designed to improve the reliability of spectral values at all natural periods by removing noise with modern proposed approaches. In this paper, a new global GMPE is proposed by using selected ground motions from SIMBAD to improve the reliability of computed spectral shape indicators. To determine regression coefficients, 204 pairs of horizontal components from 35 earthquakes with magnitude ranging from Mw 5 to Mw 7.1 and epicentral distances lower than 40 km selected from SIMBAD are used. The proposed equation is compared with similar models both qualitatively and quantitatively. After the verification of model by several goodness-of-fit measures, the epsilon values as the spectral shape indicator are computed and the validity of available prediction equations for correlation of the pairs of epsilon values is examined. General consistency between predictions by new model and others, especially, in short periods is confirmed, while, at longer periods, there are meaningful differences between normalized residuals and correlation coefficients between pairs of them estimated by new model and those are computed by other empirical equations. A simple collapse assessment example indicate possible improvement in the correlation between collapse capacity and spectral shape indicators (${\varepsilon}$) up to 20% by selection of a more applicable GMPE for calculation of ${\varepsilon}$.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.14 no.4
• /
• pp.247-256
• /
• 2021

In this paper, we propose a packet loss concealment (PLC) algorithm using pitch harmonic motion prediction and adaptive signal amplitude prediction and. The spectral motion prediction method divides the spectral motion of the previous usable frame into predetermined sub-bands to predict and restore the motion of the lost signal. In the proposed algorithm, the speech signal is classified into voiced and unvoiced sounds. In the case of voiced sounds, it is further divided into pitch harmonics using the pitch frequency to predict and restore the pitch harmonic motion of the lost frame, and for the unvoiced sound, the lost frame is restored using the spectral motion prediction method. When the continuous loss of speech frames occurs, a method of adjusting the gain using the least mean square (LMS) predictor is proposed. The performance of the proposed algorithm was evaluated through the objective evaluation method, PESQ (Perceptual Evaluation of Speech Quality) and was showed MOS 0.1 improvement over the conventional method.

• Computers and Concrete
• /
• v.22 no.2
• /
• pp.209-220
• /
• 2018

In order to evaluate the effect of hybrid fibers on the flexural performance of tunnel segment at room temperature, twelve reinforced self-consolidating concrete (SCC) symmetric inclination beams containing steel fiber, macro polypropylene fiber, micro polypropylene fiber, and their hybridizations were studied under combined loading of flexure and axial compression. The results indicate that the addition of mono steel fiber and hybrid fibers can enhance the ultimate bearing capacity and cracking behavior of tested beams. These improvements can be further enhanced along with increasing the content of steel fiber and macro PP fiber, but reduced with the increase of the reinforcement ratio of beams. The hybrid effect of steel fiber and macro PP fiber was the most obvious. However, the addition of micro PP fibers led to a degradation to the flexural performance of reinforced beams at room temperature. Meanwhile, the hybrid use of steel fiber and micro polypropylene fiber didn't present an obvious improvement to SCC beams. Compared to micro polypropylene fiber, the macro polypropylene fiber plays a more prominent role on affecting the structural behavior of SCC beams. A calculation method for ultimate bearing capacity of flexural SCC symmetric inclination beams at room temperature by taking appropriate effect of hybrid fibers into consideration was proposed. The prediction results using the proposed model are compared with the experimental data in this study and other literature. The results indicate that the proposed model can estimate the ultimate bearing capacity of SCC symmetric inclination beams containing hybrid fibers subjected to combined action of flexure and axial compression at room temperature.

• Journal of Advanced Navigation Technology
• /
• v.19 no.4
• /
• pp.318-322
• /
• 2015

In wireless communication networks, most of the prediction techniques are used for predicting the amount of resource required by user's calls for improving their demanding quality of service. However, we propose a channel allocation scheme based on predicting the resources of white spectrum holes for improving the QoS of rental user's spectrum handoff calls for cognitive radio networks in this paper. This method is supported by Wiener predictor to predict the amount of white spectrum holes of license user's free spectrum resources. We classify rental user's calls into initial calls and spectrum handoff calls, and some portion of predicted spectrum-hole resources is reserved for spectrum handoff calls' priority allocation. Simulations show that the performance of the proposed scheme outperforms in spectrum handoff call's dropping rate than an existing method without spectrum hole prediction(11% average improvement in 50% reservation).