• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.6
• /
• pp.1540-1561
• /
• 2024

This research proposes a novel approach for Tamil Handwritten Character Recognition (THCR) that combines feature selection and ensemble learning techniques. The Tamil script is complex and highly variable, requiring a robust and accurate recognition system. Feature selection is used to reduce dimensionality while preserving discriminative features, improving classification performance and reducing computational complexity. Several feature selection methods are compared, and individual classifiers (support vector machines, neural networks, and decision trees) are evaluated through extensive experiments. Ensemble learning techniques such as bagging, and boosting are employed to leverage the strengths of multiple classifiers and enhance recognition accuracy. The proposed approach is evaluated on the HP Labs Dataset, achieving an impressive 95.56% accuracy using an ensemble learning framework based on support vector machines. The dataset consists of 82,928 samples with 247 distinct classes, contributed by 500 participants from Tamil Nadu. It includes 40,000 characters with 500 user variations. The results surpass or rival existing methods, demonstrating the effectiveness of the approach. The research also offers insights for developing advanced recognition systems for other complex scripts. Future investigations could explore the integration of deep learning techniques and the extension of the proposed approach to other Indic scripts and languages, advancing the field of handwritten character recognition.

• Korean Journal of Computational Design and Engineering
• /
• v.3 no.2
• /
• pp.121-126
• /
• 1998

It is one of important issues to find intersection curve? in representation of complex surfaces on a computer. Three typical methods, i.e. the tracing method, the subdivision method, and hybrid method, are often applied to find intersection curves between sculptured surfaces. In this paper two topics are dealt with for efficiency and robustness of the hybrid method. One tropic is about how to determine step sizes variably during tracing, the ethel is about how to find tangential points between surfaces. Tracing by variable step size finds intersections rapidly and requires less memory size. Some illustrations show tangential points between surfaces.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.15 no.5
• /
• pp.705-711
• /
• 2012

The purpose of this study is to develop the model to estimate the penetration rate of metal under a high power continuous wave laser irradiation. To estimate it, an empirical modeling is more practical when the penetration phenomena of metal by laser irradiation is too complex to be analyzed by the numerical simulation. When several methods published earlier were applied to our results, we found out that their methods were not appropriate as the model. Therefore, we suggested the new empirical method considering effective intensity as a key variable. As a result, we confirmed that the new method was effective to model the penetration rate of SUS304 metal and expected that it could be available to other metals.

• Journal of IKEEE
• /
• v.28 no.3
• /
• pp.451-457
• /
• 2024

This paper proposes a method for predicting agricultural product prices by utilizing various variables such as price, climate factors, demand, and import volume as data, and applying the Long Short-Term Memory (LSTM) model. The analysis of prediction performance using the LSTM model, which learns the long-term dependencies of time series data, showed that integrating diverse data improved performance compared to traditional methods. Furthermore, even when predicting without price data as a dependent variable, meaningful results were achieved using only independent variables, indicating the potential for further model development. Moreover, it was found that using a multi-variable model could further enhance prediction performance, suggesting that this complex approach is effective in improving the accuracy of cabbage price predictions.

• Journal of Preventive Medicine and Public Health
• /
• v.50 no.5
• /
• pp.320-327
• /
• 2017

Objectives: The purpose of our study was to evaluate the association between the intensity of physical activity (PA) and prevalence of cardiovascular disease (CVD) using Korean representative data. Methods: We analyzed 39 804 participant data from the Korea National Health and Nutrition Examination Survey, 2007-2013. Exposure variable was three levels of PA (low, medium, and high) in a week, and outcome variable was prevalence of CVD based on patient self-recognition and doctor's diagnosis. Complex logistic regression analysis was performed to evaluate the relationship between level of PA and CVD adjusted by body mass index, hypertension, hypercholesterolemia, diabetes mellitus, stress recognition, household income, smoking, and current drinking. The indices of association w ere estimated as crude prevalence odds ratio (POR), adjusted POR, and their 95% confidence interval (CI). All statistical analyzes were performed using complex sample analysis procedure of the SPSS version 23.0. Results: When all variables were adjusted, only high level PA in women showed a significant association with stroke (adjusted POR by patient's self-recognition, 0.57; 95% CI, 0.32 to 0.99, adjusted POR by doctor's diagnosis, 0.55; 95% CI, 0.34 to 0.87) and CVD (adjusted POR by doctor's diagnosis, 0.68; 95% CI, 0.48 to 0.96). Conclusions: High level PA in women has a significant reverse association with prevalence of stroke and CVD in Korea. Further study for elucidating the mechanism will be needed.

• Journal of IKEEE
• /
• v.22 no.4
• /
• pp.999-1005
• /
• 2018

This paper proposed a search-based high resolution frequency estimation method in power systme. The proposed frequency estimation method adopts a slope-based adaptive search as a base of adaptive estimation structure. The architectural and operational parameters in this adaptive algorithm are changed using the information from context layer analysis of the signals including a localized full-search of spectral peak. The convergence rate of the proposed algorithm becomes much faster than those of other conventional slope-based adaptive algorithms by effectively reducing search range with the application of the localized full-search of spectrum peak. The improvements in accuracy and convergence rate of the proposed algorithm are confirmed through the performance comparison with other representative frequency estimation methods, such as, DFT(discrete Fourier transform) method, ECKF(extended complex Kalman filter), and MV(minimum variable) method.

• Journal of the korean Society of Automotive Engineers
• /
• v.25 no.4
• /
• pp.74-82
• /
• 2003

Future emission legislation requirements especially the need for CO$_2$ reduction lead to more complex powertrain concepts with an increasing number of independent parameters to be calibrated. For gasoline engines concepts with variable valve timing, direct injection or variable charge motion are in development or already on production. Diesel engines with common rail systems offer a wide range of new injection strategies, the application of new exhaust aftertreatment systems leads to additional complexity. Furthermore a clear trend to highly sophisticated transmission concepts requires a perfect interaction of all powertrain components. While the higher complexity requires increasing test and development effort, the development duration is reduced significantly. Consequently, the potential of such systems cannot be fully utilised by traditional development and calibration approaches within the given timeframe. By introduction of intelligent methodologies f3r the calibration of modem powertrains the development becomes more efficient, faster and better in quality. However, even with standardised and automated calibration methods a differentiated brand-specific powertrain character has to be maintained comparable to a "handmade" calibration performed by highly experienced experts.

• BMB Reports
• /
• v.42 no.11
• /
• pp.731-736
• /
• 2009

The generation of functional recombinant antibodies from hybridomas is necessary for antibody engineering. However, this is not easily accomplished due to high levels of aberrant heavy and light chain mRNAs, which require a highly selective technology that has proven complicated and difficult to operate. Herein, we attempt to use an alkaline phosphate (AP)-fused form of single-chain variable fragment (scFv) for the simple identification of a hybridoma-derived, functional recombinant antibody. As a representative example, we cloned the scFv gene from a hybridoma-producing mouse IgG against branched-chain keto acid dehydrogenase complex-E2 (BCKD-E2) into an expression vector containing an in-frame phoA gene. Functional recombinant antibodies were easily identified by conventional enzyme-linked immunosorbent assay (ELISA) by employing scFv-AP fusion protein, which also readily serves as a valuable immuno-detective reagent.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.6 no.8
• /
• pp.63-71
• /
• 1983

Aggregate planning coordinate the control variable over long-term to apply a demand variable and forcasting. In order to necessary the goal that doesn't make an inter-contradiction and explicitly defined. We made a considerable point of system approach for scheduling establishment. It include the control variables of aggregate planning : 1) employment 2) over time working and idle time 3) inventory 4) delivery delay S) subcontract 61 long - term facility capacity. Each variables composed of pure strategy as like a decision of inventory level, a change of employment level, etc. md alternative costs make a computation on the economic foundation. But the optimum alternative costs represent the mixed pure strategy. The faults of this method doesn't optimum guarantee a special scheduling as well as increasing a number of alternative combination. Theoretical, Linear Decision Rule make an including all variables, but it is almost impossible for this model to develope actually And also make use of the aggregate planning problem for developing system approach : LDR, heuristic model, Search Decision Rule, all kind of computers, simulation. But these models are very complex, each variables get an extremely inter-dependence. So this study be remained by theory level, some approach methods has not been brought the optimum solution to apply in every cases.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.43 no.5 s.311
• /
• pp.52-61
• /
• 2006

Orthogonal wavelet tansform which is generally used in image and signal processing applications has limited performance because of lack of shift invariance and low directional selectivity. To overcome these demerits complex wavelet transform has been proposed. In this paper, we present an efficient image denoising method using dual-tree complex wavelet transform and Bernoulli-Gauss prior model. In estimating hyper-parameters for Bernoulli-Gaussian model, we present two simple and non-iterative methods. We use hypothesis-testing technique in order to estimate the mixing parameter, Bernoulli random variable. Based on the estimated mixing parameter, variance for clean signal is obtained by using maximum generalized marginal likelihood (MGML) estimator. We simulate our denoising method using dual-tree complex wavelet and compare our algorithm to well blown denoising schemes. Experimental results show that the proposed method can generate good denoising results for high frequency image with low computational cost.