• International journal of advanced smart convergence
• /
• v.7 no.2
• /
• pp.95-100
• /
• 2018

In this paper, we propose a binarized multi-scale module to accelerate the speed of the pose estimating deep neural network. Recently, deep learning is also used for fine-tuned tasks such as pose estimation. One of the best performing pose estimation methods is based on the usage of two neural networks where one computes the heat maps of the body parts and the other computes the part affinity fields between the body parts. However, the convolution filtering with a large kernel filter takes much time in this model. To accelerate the speed in this model, we propose to change the large kernel filters with binarized multi-scale modules. The large receptive field is captured by the multi-scale structure which also prevents the dropdown of the accuracy in the binarized module. The computation cost and number of parameters becomes small which results in increased speed performance.

• Journal of KIISE
• /
• v.41 no.9
• /
• pp.686-698
• /
• 2014

Recently, large-scale streaming sensor data have emerged due to explosive supply of smart phones, diffusion of IoT and Cloud computing technology, and generalization of IoT devices. Also, researches on combination of semantic web technology are being actively pushed forward by increasing of requirements for creating new value of data through data sharing and mash-up in large-scale environments. However, we are faced with big issues due to large-scale and streaming data in the inference field for creating a new knowledge. For this reason, we propose the RDFS rule based parallel reasoning scheme to service by processing large-scale streaming sensor data with the semantic web technology. In the proposed scheme, we run in parallel each job of Rete network algorithm, the existing rule inference algorithm and sharing data using the HBase, a hadoop database, as a public storage. To achieve this, we implement our system and evaluate performance through the AWS data of the weather center as large-scale streaming sensor data.

• Journal of Power Electronics
• /
• v.18 no.2
• /
• pp.547-557
• /
• 2018

This work proposes a control method of frequency stabilization for grid integration of large-scale wind farms via the voltage source converter-based high-voltage direct current (VSC-HVDC) technology. First, the topology of grid integration of a large-scale wind farm via the VSC-HVDC link is provided, and simple control strategies for wind turbines, wind farm side VSC (WFVSC), and grid side VSC are presented. Second, a mathematical model between the phase angle of WFVSC and the frequency of the wind farm is established. The control principle of the large-scale wind power integrated system is analyzed in theory in accordance with the mathematical model. Third, frequency and AC voltage controllers of WFVSC are designed based on the mathematical model of the relationships between the phase angle of WFVSC and the frequency of the wind farm, and between the modulation index of WFVSC and the voltage of the wind farm. Corresponding controller structures are established by deriving a transfer function, and an optimization method for selecting the parameters of the frequency controller is presented. Finally, a case study is performed under different operating conditions by using the DIgSILENT/PowerFactory software. Results show that the proposed control method has good performance in the frequency stabilization of the large-scale wind power integrated system via the VSC-HVDC technology.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.20 no.2
• /
• pp.89-101
• /
• 2017

The study used GIS methods to analyze a business district consisting of traditional markets and large-scale stores, to determine the level of support needed for small enterprises in an urban area of Changwon-si, Gyeongsangnam-do. Data gathered on the area was analyzed using GIS tools such as Kernel density, Network analysis, and Huff modeling. Traditional markets are concentrated in areas where large-scale stores are located, and data analyses show that the number of consumer'use of large-scale stores (157,071) was three times that of traditional markets (59,953). One explanation for these results is that the large-scale stores are located either in densely populated areas or are adjacent to the traditional markets. Therefore, standards and regulations are needed to support small enterprise business districts. In the future, the results of this study can be used as a reference for planning and supporting traditional market business districts.

• The Journal of Fisheries Business Administration
• /
• v.46 no.3
• /
• pp.143-161
• /
• 2015

This study presents a comparative analysis on mackerel distribution process and price formation process, and investigation of price and margin between traditional markets and Large-scale discount store distribution channel. Through this, the study investigated distribution efficiency of each channel, and examined whether a difference of distribution efficiency leads to a difference of performance through the investigation of a difference of function and role between members of a wholesale market and vendor of Large scale discount store. The following are the results of this study. As a consequence of investigating supply and sum by distribution channel of mackerel, it appeared that mackerels shipped from port market are distributed into 9 consumption sites(Wholesale market, Large scale discount store, Institutional Food Service, etc.). In the comparison of distribution efficiency between traditional retail store and Large scale discount store 52.0% margin is formed in traditional retail store distribution channel and 43.1% margin is formed in Large scale discount store, and a distribution cost rate consists of 19.4% cost in a traditional retail store for fishery products and 18.1% cost in a Large-scale discount store. To analyze a difference of performance, the study examine a difference of role and function between vendor and Wholesale market company, wholesaler and middleman. Wholesale market company and middleman of wholesale market for consumer have slightly high or similar score in collection function, sorting function, evaluation function and financial function which are traditional and original. However, it was confirmed that vendor has a better score in other functions, that is, newly-demanded functions(ex : market frontier function, product development function, Integral Distribution Function, etc.).

• Fire Science and Engineering
• /
• v.22 no.1
• /
• pp.37-44
• /
• 2008

The reduced scale fire test provides basic data but it is not enough to analysis real fire problem directly because there is no exact analogy theory between a real fire and the reduced scale model. Therefore, we have developed the 10 MW large scale calorimeter in order to real scale fire test. This advanced large scale calorimeter used for physical properties such as a heat release rate, based upon consumption of $O_2$ method. Using the heptane pool fire, we carried out the calibration in order to evaluation for heat release rate. It is approve that this facility has the reliability and it is capable of applying to the advance fire research in the future.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.22 no.1
• /
• pp.34-49
• /
• 1998

In this study, turbulent flows in a grooved channel are numerically investigated by Large Eddy Simulation (LES). Especially, a parametric study is carried out to study effects of length and depth of a groove on large-scale flow structures. For one test case, comparison of LES results with those of DNS reveals a good agreement even though the number of grid points of LES is only 6.5% of that of DNS. This confirms that LES is a suitable tool for a parametric study of turbulent flows. The subsequent parametric study using LES shows that the large-scale turbulent structures are significantly affected by the geometry of the groove. Especially, when the length of the groove is short such that the recirculation region occupies the entire groove, the turbulent flow in the groove becomes very weak in both mean and fluctuation quantities.

• Proceedings of the Korean Geotechical Society Conference
• /
• 2009.03a
• /
• pp.626-633
• /
• 2009

A large scale resonant column testing equipment is under development. The would-be equipment is aiming to test 150mm-diameter specimens, which can contain as large a grain size as 25mm. Such a large specimen is hardly excitated with the existing fixed-free end condition because the torsional force cannot be effectively coupled to the specimen. The specimen will be rather resonated with free-free condition and the scheme is implemented with a rotational bearing installed between coil-magnet exciter and base pedestal. Presently the equipment was assembled and is under calibration with a cylindrical brass specimen.

• Electronics and Telecommunications Trends
• /
• v.38 no.5
• /
• pp.23-33
• /
• 2023

With the widespread demand from data-intensive tasks such as machine learning and large-scale databases, the amount of data processed in modern computing systems is increasing exponentially. Such data-intensive tasks require large amounts of memory to rapidly process and analyze massive data. However, existing computing system architectures face challenges when building large-scale memory owing to various structural issues such as CPU specifications. Moreover, large-scale memory may cause problems including memory overprovisioning. The Compute Express Link (CXL) allows computing nodes to use large amounts of memory while mitigating related problems. Hence, CXL is attracting great attention in industry and academia. We describe the overarching concepts underlying CXL and explore recent research trends in this technology.

• KSBB Journal
• /
• v.16 no.1
• /
• pp.11-14
• /
• 2001

Techniques for protein refolding from inclusion body are discussed in view of its engineering application to large scale protein purification. Among the techniques, dilution and dialysis are mainly utilized due to simple operation. Membrane reactor, gel filtration chromatography, and continuous tank operation are emerging tools for their process-scale possibility in refolding. Reaction engineering approaches could be used to analyze the kinetic behaviour in the process scale refolding reactor. The kinetic analysis is helpful in the optimization of refolding yield in the refolding reactor.