• East Asian mathematical journal
• /
• v.40 no.3
• /
• pp.307-318
• /
• 2024

This paper explores the application of matrix factorization, specifically CUR decomposition, in the clustering of Korean language documents by topic. It addresses the unique challenges of Natural Language Processing (NLP) in dealing with the Korean language's distinctive features, such as agglutinative words and morphological ambiguity. The study compares the effectiveness of Latent Semantic Analysis (LSA) using CUR decomposition with the classical Singular Value Decomposition (SVD) method in the context of Korean text. Experiments are conducted using Korean Wikipedia documents and newspaper data, providing insight into the accuracy and efficiency of these techniques. The findings demonstrate the potential of CUR decomposition to improve the accuracy of document clustering in Korean, offering a valuable approach to text mining and information retrieval in agglutinative languages.

• Journal of the Institute of Convergence Signal Processing
• /
• v.12 no.3
• /
• pp.163-168
• /
• 2011

In this paper, we proposed non-linear gamma curve algorithm for gamma correction. The previous non-linear gamma curve algorithm is generated by the least square polynomial using the Gauss-Jordan inverse matrix. However, the previous algorithm has some weak points. When calculating coefficients using inverse matrix of higher degree, occurred truncation errors. Also, only if input sample points are existed regular interval on 10-bit scale, the least square polynomial is accurately works. To compensate weak-points, we calculated accurate coefficients of polynomial using eigenvalue and orthogonal value of mat11x from singular value decomposition (SVD) and QR decomposition of vandemond matrix. Also, we used input data part segmentation, then we performed polynomial curve fitting and merged curve fitting results. When compared the previous method and proposed method using the mean square error (MSE) and the standard deviation (STD), the proposed segmented polynomial curve fitting is highly accuracy that MSE under the least significant bit (LSB) error range is approximately $10^{-9}$ and STD is about $10^{-5}$.

• Journal of Information Processing Systems
• /
• v.12 no.1
• /
• pp.46-56
• /
• 2016

Dynamic textures are videos that exhibit a stationary property with respect to time (i.e., they have patterns that repeat themselves over a large number of frames). These patterns can easily be tracked by a linear dynamic system. In this paper, a model that identifies the underlying linear dynamic system using wavelet coefficients, rather than a raw sequence, is proposed. Content based threshold filtering based on Set Partitioning in a Hierarchical Tree (SPIHT) helps to get another representation of the same frames that only have low frequency components. The main idea of this paper is to apply SPIHT based threshold filtering on different bands of wavelet transform so as to have more significant information in fewer parameters for singular value decomposition (SVD). In this case, more flexibility is given for the component selection, as SVD is independently applied to the different bands of frames of a dynamic texture. To minimize the time complexity, the proposed model is implemented on a graphics processing unit (GPU). Test results show that the proposed dynamic system, along with a discrete wavelet and SPIHT, achieve a highly compact model with better visual quality, than the available LDS, Fourier descriptor model, and higher-order SVD (HOSVD).

• Journal of Sensor Science and Technology
• /
• v.8 no.1
• /
• pp.63-68
• /
• 1999

We proposed a method of eye pattern detection in the 2-D image which was obtained by CCD video camera. To detect face region and eye pattern, we proposed pattern search network and batch SVD algorithm which had the statistical equivalence of PCA. We also used HMM to improve the accuracy of detection. As a result, we acknowledged that the proposed algorithm was superior to PCA pattern detection algorithm in computational cost and accuracy of defection. Furthermore, we evaluated that the proposed algorithm was possible in real-time face pattern detection with 2 frame images per second.

• ETRI Journal
• /
• v.28 no.1
• /
• pp.59-66
• /
• 2006

Learning behaviors of a radial basis function network (RBFN) using a singular value decomposition (SVD) and stochastic gradient (SG) algorithm, together named RBF-SVD-SG, for odor sensing systems are analyzed, and a fast training method is proposed. RBF input data is from a conducting polymer sensor array. It is revealed in this paper that the SG algorithm for the fine-tuning of centers and widths still shows ill-behaving learning results when a sufficiently small convergence coefficient is not used. Since the tuning of centers in RBFN plays a dominant role in the performance of RBFN odor sensing systems, our analysis is focused on the center-gradient variance of the RBFN-SVD-SG algorithm. We found analytically that the steadystate weight fluctuation and large values of a convergence coefficient can lead to an increase in variance of the center-gradient estimate. Based on this analysis, we propose to use the least mean square algorithm instead of SVD in adjusting the weight for stable steady-state weight behavior. Experimental results of the proposed algorithm have shown faster learning speed and better classification performance.

• Proceedings of the IEEK Conference
• /
• 2002.07c
• /
• pp.1784-1787
• /
• 2002

Two-dimensional (2-D) digital filters are widely useful inn image processing and other 2-D digital signal processing fields, but designing 2-D filters is much more difficult than designing one-dimensional (1-D) ones. This paper provides a new insight into the existing singular value decomposition (SVD)-based design approach in the sense that the SVD-based design can be performed more efficiently by exploiting the symmetries of the given 2-D magnitude specifications. By using the specification symmetries. only half of the 1-D filters (sub-filters) need to be designed. which significantly simplifies the design process and reduces the computer storage required for 1-D sub-filter coefficients. Another novel point of this paper si that an objective criterion is proposed for selecting appropriate sub-filter orders in order to reduce the hardware implementation cost. A design example is given to illustrate the effectiveness of the SVD-based design approach by exploiting specification symmetry and new order-selecting criterion.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.21 no.8
• /
• pp.1311-1321
• /
• 1997

In multibody dynamics, differential and algebraic equations which can satisfy both equation of motion and kinematic constraint equation should be solved. To solve these equations, coordinate partitioning method and constraint stabilization method are commonly used. In the coordinate partitioning method, the coordinates are divided into independent and dependent and coordinates. The most typical coordinate partitioning method are LU decomposition, QR decomposition, and SVD (singular value decomposition). The objective of this research is to find an efficient coordinate partitioning method in the dynamic analysis of flexible multibody systems. Comparing two coordinate partitioning methods, i.e. LU and QR decomposition in the flexible multibody systems, a new hybrid coordinate partitioning method is suggested for the flexible multibody analysis.

• The Journal of the Petrological Society of Korea
• /
• v.4 no.1
• /
• pp.49-58
• /
• 1995

Staurolite-biotite-andalusite-garnet (SBAG) assemblage and its sub-assemblages (SBA and SBG) commonly occur in the Dueumri Formation near the Chunyang granite, belonging to andalusite and sillimanite zones. The occurrence of the SBAG mineral assemblage is unusual because it is univariant in the $K_2O-FeO-MgO-Al_2O_3-SiO_2-H_2O$ (KFMASH) model system. We used projection and singular value decomposition (SVD) methods to investigate the equilibrium relationship between SBAG and its sub-assemblage. The SVD modelling of single specimen containing the SBAG assemblage suggests no reaction relationship with respect to mass-balance. Thus, the SBAG assemblages are stabilized by non-KFMASH component. On the other hand, the AFM-Mn projection suggests a reaction relationship between SBAG and its sub-assemblage because they intersect each other in this composition space. The SVD modelling, however, suggests no reaction relationship between these assemblages. Thus, the SBAG assemblages are likely to be stabilized by the variation in bulk-rock composition and/or 1.1~2,. The stable occurrence of staurolite in the sillimanite zone is compatible with pressure estimates from the garnet-plagioclase-biotite-muscovite geobarometer.

• Journal of Internet Computing and Services
• /
• v.25 no.1
• /
• pp.109-121
• /
• 2024

In various manufacturing processes such as textiles and automobiles, when equipment breaks down or stops, the machines do not work, which leads to time and financial losses for the company. Therefore, it is important to detect equipment abnormalities in advance so that equipment failures can be predicted and repaired before they occur. Most equipment failures are caused by bearing failures, which are essential parts of equipment, and detection bearing anomaly is the essence of PHM(Prognostics and Health Management) research. In this paper, we propose a preprocessing algorithm called SWT-SVD, which analyzes vibration signals from bearings and apply it to an anomaly transformer, one of the time series anomaly detection model networks, to implement bearing anomaly detection model. Vibration signals from the bearing manufacturing process contain noise due to the real-time generation of sensor values. To reduce noise in vibration signals, we use the Stationary Wavelet Transform to extract frequency components and perform preprocessing to extract meaningful features through the Singular Value Decomposition algorithm. For experimental validation of the proposed SWT-SVD preprocessing method in the bearing anomaly detection model, we utilize the PHM-2012-Challenge dataset provided by the IEEE PHM Conference. The experimental results demonstrate significant performance with an accuracy of 0.98 and an F1-Score of 0.97. Additionally, to substantiate performance improvement, we conduct a comparative analysis with previous studies, confirming that the proposed preprocessing method outperforms previous preprocessing methods in terms of performance.

• Journal of Biomedical Engineering Research
• /
• v.25 no.6
• /
• pp.589-598
• /
• 2004

We propose a new algorithm to extract the fetal electrocardiogram (FECG) from a multi-channel electrocardiogram (ECG) recorded at the chest and abdomen of a pregnant woman. To extract the FECG from the composite abdominal ECG, the classical time-domain method based on singular value decomposition (SVD) has been generally used. However, this method has some disadvantages, such as its high degree of computational complexity and the necessary assumption that vectors between the FECG and the maternal electrocardiogram (MECG) should be orthogonal. The proposed algorithm, which uses SVD in a discrete cosine transform (DCT) domain, compensates for these disadvantages. To perform SVD with lower computational complexity, DCT coefficients corresponding to high-frequency components were eliminated on the basis of the properties of the DCT coefficients and the frequency characteristics of the FECG. Moreover, to extract the pure FECG with little influence of the direction of the vectors between the FECG and MECG, three new channels were made out of the MECG suppressed in the composite abdominal ECG, and the new channels were appended to the original multi-channel ECG. The performance of the proposed algorithm and the classical time-domain method based on SVD were compared using simulated and real data. It was experimentally verified that the proposed algorithm can extract the pure FECG with reduced computational complexity.