• Journal of the Semiconductor & Display Technology
• /
• v.10 no.3
• /
• pp.75-81
• /
• 2011

This paper presents a novel method for image classification based on a hybrid genetic algorithm (GA) and support vector machine (SVM) approach which can significantly improve the classification performance for content-based image retrieval (CBIR). Though SVM has been widely applied to CBIR, it has some problems such as the kernel parameters setting and feature subset selection of SVM which impact the classification accuracy in the learning process. This study aims at simultaneously optimizing the parameters of SVM and feature subset without degrading the classification accuracy of SVM using GA for CBIR. Using the hybrid GA and SVM model, we can classify more images in the database effectively. Experiments were carried out on a large-size database of images and experiment results show that the classification accuracy of conventional SVM may be improved significantly by using the proposed model. We also found that the proposed model outperformed all the other models such as neural network and typical SVM models.

• Journal of Electrical Engineering and Technology
• /
• v.7 no.6
• /
• pp.948-953
• /
• 2012

The characteristics of an induction motor vary with the number of parameters and the performance relationship between the parameters also is implicit. In case of the induction motor design, we generally should estimate many objective physical quantities in the optimization procedure. In this article, the multi objective design optimization based on genetic algorithm is applied for the three phase induction motor. The efficiency, starting torque, and material cost are selected for the objectives. The validity of the design results is also clarified by comparison between calculated results and measured ones.

• Korean Journal of Remote Sensing
• /
• v.23 no.3
• /
• pp.189-198
• /
• 2007

Introduction of lidar technology have contributed to a wide range of applications in generating quality surface models. Accordingly, because of the importance of terrain surface models in mapping applications, rigorous studies have been performed to extract ground points from a lidar data point cloud. Although most filters have been shown abilities to extract ground points with their parameters tuned, however, most experiments revealed that there are certain limitations in optimizing filter parameters and the correction of remaining misclassified points is not straightforward. In this study, therefore, a method to improve the quality of filtered lidar data is proposed, which exploits neighboring surface properties arising between immediate neighbors. The method comprises a sequence of procedures which can reduce commission and omission errors. Commission errors occurring in low-rise objects are reduced by utilizing morphological operations. On the other hand, omission errors are reduced by adding missing ground points around step edges. Experimental results show that the qualities of filtered data can be improved considerably by the proposed method.

• Korean Journal of Air-Conditioning and Refrigeration Engineering
• /
• v.11 no.5
• /
• pp.700-708
• /
• 1999

The optimum dimple shape and arrangement in the channel of a plate heat exchanger with staggered dimples are proposed in this study. Four important geometric parameters are selected as design variables, the pressure drop and heat transfer characteristics are examined in the channel of plate heat exchangers. The optimization is accomplished by minimizing the global criterion function which consists of the correlations of Nusselt number and pressure drop. The optimum geometric parameters were found at the dimensionless dimple distance (L) of 0.272, the dimensionless dimple angle ($\beta$) of 0.44, the dimensionless dimple volume (V) of 0.106 and the dimensionless dimple pitch (G) of 0.195. It is found that the heat transfer and pressure drop of the optimum model are increased by approximately 227.9% and 32.9%, respectively, compared to those of the base model.

• Smart Media Journal
• /
• v.9 no.4
• /
• pp.36-43
• /
• 2020

The Monte Carlo tree search (MCTS) is a popular method for implementing an intelligent game program. It has several hyper-parameters that require an optimization for showing the best performance. Due to the stochastic nature of the MCTS, the hyper-parameter optimization is difficult to solve. This paper uses the self-playing capability of the MCTS-based game program for optimizing the hyper-parameters. It seeks a winner path over the hyper-parameter space while performing the self-play. The top-q longest winners in the winner path compete for the final winner. The experiment using the 15-15-5 game (Omok in Korean name) showed a promising result.

• International journal of advanced smart convergence
• /
• v.13 no.2
• /
• pp.80-87
• /
• 2024

This paper focuses on improving accuracy in constrained computing settings by employing the ReLU (Rectified Linear Unit) activation function. The research conducted involves modifying parameters of the ReLU function and comparing performance in terms of accuracy and computational time. This paper specifically focuses on optimizing ReLU in the context of a Multilayer Perceptron (MLP) by determining the ideal values for features such as the dimensions of the linear layers and the learning rate (Ir). In order to optimize performance, the paper experiments with adjusting parameters like the size dimensions of linear layers and Ir values to induce the best performance outcomes. The experimental results show that using ReLU alone yielded the highest accuracy of 96.7% when the dimension sizes were 30 - 10 and the Ir value was 1. When combining ReLU with the Adam optimizer, the optimal model configuration had dimension sizes of 60 - 40 - 10, and an Ir value of 0.001, which resulted in the highest accuracy of 97.07%.

• Journal of the Semiconductor & Display Technology
• /
• v.23 no.2
• /
• pp.55-59
• /
• 2024

Plasma Enhanced Chemical Vapor Deposition (PE-CVD) is a widely used technology in semiconductor manufacturing for thin film deposition. The implementation of wafer guide rings in PE-CVD processes is crucial for enhancing efficiency and product quality by ensuring uniform deposition around wafer edges and reducing particle generation. On the other hand, to prevent overall temperature non-uniformity and degradation of thin film quality within the chamber, it is essential to consider various parameters comprehensively. In this study, after applying the wafer guide rings, temperature variations and fluid flow changes were simulated. Additionally, by simulating the temperature and flow changes when applied to the PE-CVD chamber, this paper discusses the importance of optimizing variables within the entire chamber.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.11
• /
• pp.1551-1558
• /
• 2021

Beacon signals used in indoor positioning system are reflected and distorted, resulting in noise signals. KF(Kalman Filter) has been widely used to remove this noise. In order to apply the KF, optimization process considering the signal type, signal strength, and environmental elements of each product is required. In this paper, we propose a solution to the optimization problem of KF Parameters using GA(Genetic Algorithm) in BLE(Bluetooth Low Energy) Beacon-based indoor positioning system. After optimizing KF Parameters by applying the proposed technique with a certain distance between Beacon and receiver, we compared the estimated distance passed through KF with the unfiltered distance. The proposed technique is expected to reduce the time required and improve accuracy of KF Parameters optimization in an indoor positioning system based on RSSI (Received Signal Strength Indication).

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.11
• /
• pp.3097-3107
• /
• 1999

Recently many studies have been done on the Random Early Detection(RED) algorithm as an active queue management and congestion avoidance scheme in the Internet. In this paper we first overview the characteristics of RED and the modified RED algorithms in order to understand the current status of these studies. Then we analyze the RED dynamics by investigating how RED parameters affect router queue behavior. We show the cases when RED fails since it cannot react to queue state changes aggressively due to the deterministic use of its parameters. Based on the RED parameter analysis, we propose a self-adaptive algorithm to cope with this RED weakness. In this algorithm we make two parameters be adjusted themselves depending on the queue states. One parameter is the maximum probability to drop or mark the packet at the congestion state. This parameter can be adjusted to react the long burst of traffic, consequently reducing the congestion disaster. The other parameter is the queue weight which is also adjusted aggressively in order for the average queue size to catch up with the current queue size when the queue moves from the congestion state to the stable state.

• Smart Structures and Systems
• /
• v.12 no.1
• /
• pp.1-22
• /
• 2013

Magnetic nanoparticle based bioseparation in microfluidics is a multiphysics phenomenon that involves interplay of various parameters. The ability to understand the dynamics of these parameters is a prerequisite for designing and developing more efficient magnetic cell/bio-particle separation systems. Therefore, in this work proof-of-concept experiments are combined with advanced numerical simulation to design and optimize the capturing process of magnetic nanoparticles responsible for efficient microfluidic bioseparation. A low cost generic microfluidic platform was developed using a novel micromolding method that can be done without a clean room techniques and at much lower cost and time. Parametric analysis using both experiments and theoretical predictions were performed. It was found that flow rate and magnetic field strength greatly influence the transport of magnetic nanoparticles in the microchannel and control the capturing efficiency. The results from mathematical model agree very well with experiments. The model further demonstrated that a 12% increase in capturing efficiency can be achieved by introducing of iron-grooved bar in the microfluidic setup that resulted in increase in magnetic field gradient. The numerical simulations were helpful in testing and optimizing key design parameters. Overall, this work demonstrated that a simple low cost experimental proof-of-concept setup can be synchronized with advanced numerical simulation not only to enhance the functional performance of magneto-fluidic capturing systems but also to efficiently design and develop microfluidic bioseparation systems for biomedical applications.