• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.22 no.1
• /
• pp.15-28
• /
• 2018

Recently Machine Learning algorithms are widely used to process Big Data in various applications and a lot of these applications are executed in run time. Therefore the speed of Machine Learning algorithms is a critical issue in these applications. However the most of modern iteration Machine Learning algorithms use a successive iteration technique well-known in Numerical Linear Algebra. But this technique has a very low convergence, needs a lot of iterations to get solution of considering problems and therefore a lot of time for processing even on modern multi-core computers and clusters. Tchebychev iteration technique is well-known in Numerical Linear Algebra as an attractive candidate to decrease the number of iterations in Machine Learning iteration algorithms and also to decrease the running time of these algorithms those is very important especially in run time applications. In this paper we consider the usage of Tchebychev iterations for acceleration of well-known K-Means and SVM (Support Vector Machine) clustering algorithms in Machine Leaning. Some examples of usage of our approach on modern multi-core computers under Apache Spark framework will be considered and discussed.

• Korean Journal of Air-Conditioning and Refrigeration Engineering
• /
• v.14 no.4
• /
• pp.350-357
• /
• 2002

Optimal control algorithms for the full storage ice cooling system were developed by using a dynamic simulation program. Control algorithms for the storage charging mode were developed for the chiller outlet temperature setpoint control and the chiller capacity control. Control algorithms for the storage discharging mode were developed for the proper mode selection, the storage-only mode control, and the storage-priority chiller-shared mode control. Two different cases of the expected outdoor air temperature profile and the expected cooling load profile were used to analyze the effectiveness of these algorithms. Simulation results show the energy savings and the satisfactory controls of the ice storage system. Therefore, control algorithms developed for this study may effectively be used for the improved control of the ice storage cooling system.

• Journal of Institute of Control, Robotics and Systems
• /
• v.15 no.10
• /
• pp.1047-1055
• /
• 2009

Since nuclear and chemical materials could damage people and disturb battlefield missions in a wide region, nuclear/chemical reconnaissance systems utilizing multiple mobile robots are highly desirable for rapid and safe reconnaissance. In this paper, we design a nuclear/chemical reconnaissance system including mobile robots. Also we propose time-efficient trajectory planning algorithms using grid coverage and contour finding methods for reconnaissance operation. For grid coverage, we performed in analysis on time consumption for various trajectory patterns generated by straight lines and arcs. We proposed BCF (Bounded Contour Finding) and BCFEP (Bounded Contour Finding with Ellipse Prediction) algorithms for contour finding. With these grid coverage and contour finding algorithms, we suggest trajectory planning algorithms for single, two or four mobile robots. Various simulations reveal that the proposed algorithms improve time-efficiency in nuclear/chemical reconnaissance missions in the given area. Also we conduct basic experiments using a commercial mobile robot and verify the time efficiency of the proposed contour finding algorithms.

• ETRI Journal
• /
• v.38 no.3
• /
• pp.540-550
• /
• 2016

Spectral clustering is a powerful tool for exploratory data analysis. Many existing spectral clustering algorithms typically measure the similarity by using a Gaussian kernel function or an undirected k-nearest neighbor (kNN) graph, which cannot reveal the real clusters when the data are not well separated. In this paper, to improve the spectral clustering, we consider a robust similarity measure based on the shared nearest neighbors in a directed kNN graph. We propose two novel algorithms for spectral clustering: one based on the number of shared nearest neighbors, and one based on their closeness. The proposed algorithms are able to explore the underlying similarity relationships between data points, and are robust to datasets that are not well separated. Moreover, the proposed algorithms have only one parameter, k. We evaluated the proposed algorithms using synthetic and real-world datasets. The experimental results demonstrate that the proposed algorithms not only achieve a good level of performance, they also outperform the traditional spectral clustering algorithms.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2001.05a
• /
• pp.144-147
• /
• 2001

In ATM Network, many algorithms have been proposed for rate-based ABR flow control. The object of these algorithms is to turn around from cell loss and to use the unused bandwidth by ABR Traffic. These Algorithms are applied to control ABR Traffic by EFCI and ER switches. In this paper, we apply these algorithms to the various stale networks and analysis the characteristics of the algorithms and then attempt to show the loss rate of each algorithms. As the result we compare and analysis the efficiency of them.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2001.10a
• /
• pp.315-318
• /
• 2001

We propose two new blind LMS and MMSE algorithms called projection-based least mean square (PB-LMS) and projection-based minimum mean square error (PB-MMSE) for smart antennas. Both algorithms employ the finite constellation property of digital signal to transform the conventional LMS and MMSE algorithms into blind algorithms. Computer simulations were carried out in the AWGN channel and Rayleigh fading channel with AWGN in CDMA environment to verify the performance of the two proposed algorithms.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2001.10a
• /
• pp.331-335
• /
• 2001

In this paper, we propose two novel blind LMS algorithms, called exponential step sire LMS algorithms (ES-LMS), for adaptive array antennas whose convergence speed is increased, hence they are much more capable of tracking the desired signal than the conventional LMS algorithms. Both of the algorithms require neither spatial knowledge nor reference signals since they use the finite symbol property of digital signal. Computer simulations were carried cot in CDMA environment affected by multi-path Rayleigh fading to verify the performance of the two proposed algorithms.

• International Journal of Computer Science & Network Security
• /
• v.21 no.7
• /
• pp.221-232
• /
• 2021

In recent years, with the existence of huge amounts of data stored in huge databases, the need for developing accurate tools for analyzing data and extracting information and knowledge from the huge and multi-source databases have been increased. Hence, new and modern techniques have emerged that will contribute to the development of all other sciences. Knowledge discovery techniques are among these technologies, one popular technique of knowledge discovery techniques is data mining which aims to knowledge discovery from huge amounts of data. Such modern technologies of knowledge discovery will contribute to the development of all other fields. Data mining is important, interesting technique, and has many different and varied algorithms; Therefore, this paper aims to present overview of data mining, and clarify the most important of those algorithms and their uses.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.12 no.4
• /
• pp.99-114
• /
• 2002

Improved algorithms for elliptic scalar multiplication secure against side-channel attacks, such as timing and power analysis, are presented and analyzed. We first point out some potential security flaws often overlooked in most previous algorithms and then present a simple $\pm$1-signed encoding scheme that can be used to enhance the security and performance of existing algorithms. More specifically, we propose concrete signed binary and window algorithms based on the proposed $\pm$ 1-signed encoding and analyze their security and performance. The proposed algorithms are shown to be more robust and efficient than previous algorithms.

• Proceedings of the KSR Conference
• /
• 2011.10a
• /
• pp.1493-1496
• /
• 2011

Research results for LMS-based algorithms performances using real records of the traffic noise are discussed. The various algorithms based on LMS method are studied regarding their convergence speed and noise reduction index. Most effective algorithms are chosen for implementation in the active noise control barriers. The optimal step size, and number of adaptive filter taps are addressed during parametric study of the algorithms.