• Journal of the Korean Data and Information Science Society
• /
• v.25 no.3
• /
• pp.677-684
• /
• 2014

Support vector quantile regression (SVQR) is capable of providing more complete description of the linear and nonlinear relationships among random variables. To improve the estimation performance of SVQR we propose to use SVQR ensemble with bagging (bootstrap aggregating), in which SVQRs are trained independently using the training data sets sampled randomly via a bootstrap method. Then, they are aggregated to obtain the estimator of the quantile regression function using the penalized objective function composed of check functions. Experimental results are then presented, which illustrate the performance of SVQR ensemble with bagging.

• Proceedings of the KIEE Conference
• /
• 1999.07g
• /
• pp.3150-3152
• /
• 1999

The efficient representation and encoding of signals with limited resources, e.g., finite storage capacity and restricted transmission bandwidth, is a fundamental problem in technical information processing systems. Typically under realistic circumstances, the encoding and communication of message has to deal with different sources of noise and disturbances. In this paper, I propose a unifying approach to data compression by robust vector quantization, which explicitly deals with channel noise, and random elimination of prototypes. The resulting algorithm is able to limit the detrimental effect of noise in a very general communication scenario. In this paper, based on the robust vector quantization I have an experiment about speech coding.

• Communications for Statistical Applications and Methods
• /
• v.17 no.2
• /
• pp.183-191
• /
• 2010

Support vector quantile regression(SVQR) is capable of providing more complete description of the linear and nonlinear relationships among random variables. In this paper we propose an iterative reweighted least squares(IRWLS) procedure to solve the problem of SVQR with a weighted quadratic loss function. Furthermore, we introduce the generalized approximate cross validation function to select the hyperparameters which affect the performance of SVQR. Experimental results are then presented which illustrate the performance of the IRWLS procedure for SVQR.

• Korean journal of applied entomology
• /
• v.54 no.2
• /
• pp.91-97
• /
• 2015

The sweetpotato whitefly, Bemisia tabaci, is the major insect pest that transmitted over 100 plant viruses including tomato yellow leaf curl virus (TYLCV) of tomato plant as virus vector in the world. In this study, cDNA library of whitefly was constructed using Gateway system for selecting target gene in order to control of B. tabaci using virus-induced gene silencing (VIGS) vector with RNAi. First of all, when using oligo d(T) rimer, the calculated titer of cDNA library was confirmed with $1.4{\times}10^4$ clones and average insert sizes was confirmed with 1 kb. However, insert size was very big for construction of cDNA. Otherwise, when using attB-N25 random primer and sonication for 6 sec, the calculated titer of cDNA library was confirmed with $1.04{\times}10^5$ clones. But mostly insert band wasn't identified on the electrophoresis, because it seemed that insert size is too small (${\leq}100bp$), also the size of identified insert was somewhat big. Finally, when using oligo d(T) primer and sonication for 1 sec, cDNA insert of whitefly was appropriated for VIGS with 300-600 bp. However, cDNA sequence included a poly A and titer was very low to $5.2{\times}10^2$ clones. It was supposed that heat shock transformation was used instead of electro-transformation. It is considered that when constructing cDNA library for using VIGS vector, (1) random primer should be used for First strand cDNA synthesis in order to remove poly A and (2) sonication for 1 sec should be performed in order to get appropriated insert size and (3) electro-transformation should be performed in order to improve transformation efficiency.

• Journal of Environmental Science International
• /
• v.25 no.4
• /
• pp.603-613
• /
• 2016

Fog may have a significant impact on road conditions. In an attempt to improve fog predictability in Jeju, we conducted machine learning with various data mining techniques such as tree models, conditional inference tree, random forest, multinomial logistic regression, neural network and support vector machine. To validate machine learning models, the results from the simulation was compared with the fog data observed over Jeju(184 ASOS site) and Gosan(185 ASOS site). Predictive rates proposed by six data mining methods are all above 92% at two regions. Additionally, we validated the performance of machine learning models with WRF (weather research and forecasting) model meteorological outputs. We found that it is still not good enough for operational fog forecast. According to the model assesment by metrics from confusion matrix, it can be seen that the fog prediction using neural network is the most effective method.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.6
• /
• pp.239-245
• /
• 2004

A hue-based noise-tolerant corner detector is proposed for the exact detection of the real corners in spite of the shadows and random noise. Based on the fact that the hue gradient at the border of the opaque objects' shadow is smaller than the intensity gradient in HSI (hue-saturation-intensity) color space, the effects of shadow are eliminated by introducing the hue-weighted combination of vector gradient to the proposed corner detector. Furthermore, the proposed corner detector is robust to random noise by offsetting the contribution to the corner candidate when the polarities of the color gradients of the pixel pairs are out of phase each other. Results of the experiment show that the proposed corner detector can effectively detect the real corners.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.4 s.310
• /
• pp.39-49
• /
• 2006

This paper proposes an efficient verification environment of PCI Express controller that is the future I/O interface. This verification environment consists of a test vector generator, a test bench, and two abstract memories. We also define the assembler set to generate the verification scenarios. In this paper, we propose the random test environment which consists of a random vector generator, a .simulator part, and a compare engine. This verification methodology is useful to find the special errors which are not detected by the basic-behavioral test and hardware-design test.

• Korean Journal of Cognitive Science
• /
• v.21 no.4
• /
• pp.655-667
• /
• 2010

The named entity recognition task is one of the most important subtasks in Information Extraction. In this paper, we describe a Korean named entity recognition using structural Support Vector Machines (structural SVMs) and modified Pegasos algorithm. Using the proposed approach, we could achieve an 85.43% F1 and an 86.79% F1 for 15 named entity types on TV domain and sports domain, respectively. Moreover, we reduced the training time to 4% without loss of performance compared to Conditional Random Fields (CRFs).

• The Korean Journal of Applied Statistics
• /
• v.30 no.6
• /
• pp.851-865
• /
• 2017

Ransomware such as WannaCry is a global issue and methods to defend against malware attacks are important. We have to be able to classify the malware types efficiently in order to minimize the damage from malwares. This study makes models to classify malware properly with various statistical techniques. Several classification techniques such as logistic regression, random forest, gradient boosting, and support vector machine are used to construct models. This study also helps us understand key variables to classify the type of malicious software.

• Steel and Composite Structures
• /
• v.37 no.2
• /
• pp.193-209
• /
• 2020

In this paper, the efficiency of five Machine Learning (ML) methods consisting of Deep Learning (DL), Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), and Gradient Tree Booting (GTB) for regression and classification of the Ultimate Load Factor (ULF) of nonlinear inelastic steel frames is compared. For this purpose, a two-story, a six-story, and a twenty-story space frame are considered. An advanced nonlinear inelastic analysis is carried out for the steel frames to generate datasets for the training of the considered ML methods. In each dataset, the input variables are the geometric features of W-sections and the output variable is the ULF of the frame. The comparison between the five ML methods is made in terms of the mean-squared-error (MSE) for the regression models and the accuracy for the classification models, respectively. Moreover, the ULF distribution curve is calculated for each frame and the strength failure probability is estimated. It is found that the GTB method has the best efficiency in both regression and classification of ULF regardless of the number of training samples and the space frames considered.