• Journal of KIISE:Software and Applications
• /
• v.31 no.10
• /
• pp.1386-1399
• /
• 2004

This paper presents correlation-based automatic image captioning. Given a training set of annotated images, we want to discover correlations between visual features and textual features, so that we can automatically generate descriptive textual features for a new unseen image. We develop models with multiple design alternatives such as 1) adaptively clustering visual features, 2) weighting visual features and textual features, and 3) reducing dimensionality for noise sup-Pression. We experiment thoroughly on 10 data sets of various content styles from the Corel image database, about 680MB. The major contributions of this work are: (a) we show that careful weighting visual and textual features, as well as clustering visual features adaptively leads to consistent performance improvements, and (b) our proposed methods achieve a relative improvement of up to 45% on annotation accuracy over the state-of-the-art, EM approach.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.1
• /
• pp.333-342
• /
• 2012

This paper presents a vowel onset point (VOP) detection method based on the human auditory system. This method maps the "perceptual" frequency scale, i.e. Mel scale onto a linear acoustic frequency, and then establishes a series of Triangular Mel-weighted Filter Bank simulate the function of band pass filtering in human ear. This nonlinear critical-band filter bank helps greatly reduce the data dimensionality, and eliminate the effect of harmonic waves to make the formants more prominent in the nonlinear spaced Mel spectrum. The sum of mel spectrum peaks energy is extracted as feature for each frame, and the instinct at which the energy amplitude starts rising sharply is detected as VOP, by convolving with Gabor window. For the single-word database which contains 12 vowels articulated with different kinds of consonants, the experimental results showed a good average detection rate of 72.73%, higher than other vowel detection methods based on short-time energy and zero-crossing rate.

• The KIPS Transactions:PartB
• /
• v.14B no.2
• /
• pp.135-140
• /
• 2007

Feature selection is a preprocessing technique commonly used on high dimensional data. Feature selection studies how to select a subset or list of attributes that are used to construct models describing data. Feature selection methods attempt to explore data's intrinsic properties by employing statistics or information theory. The recent developments have involved approaches like correlation method, dimensionality reduction and mutual information technique. This feature selection have become the focus of much research in areas of applications with massive and complex data sets. In this paper, we provide a feature selection method considering data characteristics and generalization capability. It provides a computational approach for feature selection based on fuzzy cluster analysis of its attribute values and its performance measures. And we apply it to the system for classifying computer virus and compared with heuristic method using the contrast concept. Experimental result shows the proposed approach can give a feature ranking, select the features, and improve the system performance.

• Journal of Communications and Networks
• /
• v.13 no.3
• /
• pp.232-239
• /
• 2011

This paper investigates the user selection problem of successive zero-forcing precoded multiuser multiple-input multiple-output (MU-MIMO) downlink systems, in which the base station and mobile receivers are equipped with multiple antennas. Assuming full knowledge of the channel state information at the transmitter, dirty paper coding (DPC) is an optimal precoding strategy, but practical implementation is difficult because of its excessive complexity. As a suboptimal DPC solution, successive zero-forcing DPC (SZF-DPC) was recently proposed; it employs partial interference cancellation at the transmitter with dirty paper encoding. Because of a dimensionality constraint, the base station may select a subset of users to serve in order to maximize the total throughput. The exhaustive search algorithm is optimal; however, its computational complexity is prohibitive. In this paper, we develop two low-complexity user scheduling algorithms to maximize the sum rate capacity of MU-MIMO systems with SZF-DPC. Both algorithms add one user at a time. The first algorithm selects the user with the maximum product of the maximum column norm and maximum eigenvalue. The second algorithm selects the user with the maximum product of the minimum column norm and minimum eigenvalue. Simulation results demonstrate that the second algorithm achieves a performance similar to that of a previously proposed capacity-based selection algorithm at a high signal-to-noise (SNR), and the first algorithm achieves performance very similar to that of a capacity-based algorithm at a low SNR, but both do so with much lower complexity.

• Korean Journal of Cognitive Science
• /
• v.14 no.1
• /
• pp.10-10
• /
• 2003

This paper extracts the edge of main components of face with Gabor wavelets transformation in facial expression images. FCM(Fuzzy C-Means) clustering algorithm then extracts the representative feature points of low dimensionality from the edge extracted in neutral face. The feature-points of the neutral face is used as a template to extract the feature-points of facial expression images. To match point to Point feature points on an expression face against each feature point on a neutral face, it consists of two steps using a dynamic linking model, which are called the coarse mapping and the fine mapping. This paper presents an automatic extraction of feature-points by dynamic linking model based on Gabor wavelets and fuzzy C-means(FCM) algorithm. The result of this study was applied to extract features automatically in facial expression recognition based on dimension[1].

• Journal of the Korean Society for Marine Environment & Energy
• /
• v.16 no.2
• /
• pp.61-70
• /
• 2013

Hydrodynamic aspects on three-dimensional effects were investigated in this study for simple and convenient conversion of tidal stream energy using a Vertical-Axis Turbine (VAT). Numerical approach was made to reveal the differences of flow physics between 2-D estimation and rigorous 3-D simulation. It was shown that the 3-D effects were dominant mainly due to the variation of tip vortices around the tip region of rotor blade, causing the loss of lift for steadily translating hydrofoil and the reduction of torque for rotating turbine blade. The 3-D effect was found to be rather prominent for the typical VATs considered in this paper. Simple and yet efficient 2-D approach with the correction of its three-dimensionality was also proposed for practical design and analysis of VAT.

• Journal of Biomedical Engineering Research
• /
• v.37 no.2
• /
• pp.61-67
• /
• 2016

This paper presented a method for random forest based the arrhythmia classification using both heart rate (HR) and heart rate variability (HRV) features. We analyzed the MIT-BIH arrhythmia database which contains half-hour ECG recorded from 48 subjects. This study included not only the linear features but also non-linear features for the improvement of classification performance. We classified abnormal ECG using mean_NN (mean of heart rate), SD1/SD2 (geometrical feature of poincare HRV plot), SE (spectral entropy), pNN100 (percentage of a heart rate longer than 100 ms) affecting accurate classification among combined of linear and nonlinear features. We compared our proposed method with Neural Networks to evaluate the accuracy of the algorithm. When we used the features extracted from the HRV as an input variable for classifier, random forest used only the most contributed variable for classification unlike the neural networks. The characteristics of random forest enable the dimensionality reduction of the input variables, increase a efficiency of classifier and can be obtained faster, 11.1% higher accuracy than the neural networks.

• Journal of Internet Computing and Services
• /
• v.10 no.4
• /
• pp.1-11
• /
• 2009

Recently there has arisen concern in both the database community and the graphics society about data retrieval from large motion databases because the high dimensionality of motion data implies high costs. In this circumstance, finding an effective distance measure and an efficient query processing method for such data is a challenging problem. This paper presents an elaborate motion query processing system, SMoFinder (Similar Motion Finder), which incorporates a novel kinematic distance measure and an efficient indexing strategy via adaptive frame segmentation. To this end, we regard human motions as multi-linkage kinematics and propose the weighted Minkowski distance metric. For efficient indexing, we devise a new adaptive segmentation method that chooses representative frames among similar frames and stores chosen frames instead of all frames. For efficient search, we propose a new search method that processes k-nearest neighbors queries over only representative frames. Our experimental results show that the size of motion databases is reduced greatly (${\times}1/25$) but the search capability of SMoFinder is equal to or superior to that of other systems.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.13 no.2
• /
• pp.287-298
• /
• 1989

A numerical work was performed to study the flow behaviors of the open channel type flow with its geometric boundary conditions being similar to that of the Multi-Stage-Flash evaporator with and without a baffle. For the analysis, two-dimensional steady turbulent flow was assumed and the widely known k-.epsilon. turbulence model was usded. SIMPLE algorithm and the power difference scheme were used for the numerical approach. Numerical results generally agree with the previous experimental results though there are some uncertainties at far downstream and near the free surface due to the three dimensionality of the flow and surface waves. Without a baffle, the flow has basically the shape of the submerged plane wall jet with its upper boundary at downstream being sharply curved toward the free surface. For the flow with a baffle, recirculation flow patterns are observed at the upper inlet portion and at the backside of the baffle. For the case without a baffle, it was also confirmed that the ratio between the liquid level and the gate opening height is the most important parameter to determine the flow behavior.

• Fashion & Textile Research Journal
• /
• v.12 no.1
• /
• pp.38-45
• /
• 2010

Cosmetics retailers would benefit from studies that examine which shopping-mall attributes can be manipulated to favorably affect consumer satisfaction and revisit intention at Internet. The purposes of this study were (1) to examine the dimensionality of shopping-mall attribute for cosmetics retailers, (2) to determine which dimensions of shopping-mall attribute were significant predictors of consumer satisfaction and revisit intention and (3) to find out the moderating effect of consumer satisfaction through shopping-mall attributes on revisit intention to buy cosmetics across the types of shopping-mall at Internet (i.e., total mall and specialty mall). Data were collected from 209 online cosmetic shoppers among high school girls. Factor analysis identified five dimensions of shopping-mall attributes at Internet, such as Convenience, Price, Loading speed, Sales promotion, and Service. Only two dimensions(i.e., convenience and service) were significant predictors of online shopper satisfaction in both total mall and specialty mall. The moderating effect of consumer satisfaction on revisit intention was significant in both two mall types at Internet. For total mall, price was a significant predictor through consumer satisfaction on revisit intention, while loading speed was a significant predictor directly on revisit intention for specialty mall. In light of the major findings, this study sets forth strategic implications for consumer satisfaction and revisit intention to buy cosmetics in the setting of electronic commerce.