• Proceedings of the Korea Information Processing Society Conference
• /
• 2009.04a
• /
• pp.834-837
• /
• 2009

There is a growing need for ad-hoc analysis of extremely large data sets, especially at e-Commerce companies which depend on recommender application. Nowadays, as the number of e-Commerce web pages grow to a tremendous proportion; vertical recommender services can help customers to find what they need. Recommender application is one of the reasons for e-Commerce success in today's world. Compared with general e-Commerce recommender application, obviously, general e-Commerce recommender application's processing scope is greatly narrowed down. MapReduce is emerging as an important programming model for large-scale data-parallel applications such as web indexing, data mining, and scientific simulation. The objective of this paper is to explore MapReduce framework for the e-Commerce recommender application on major general and dedicated link analysis for e-Commerce recommender application, and thus the responding time has been decreased and the recommender application's accuracy has been improved.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.2
• /
• pp.33-42
• /
• 2022

In this paper, we proposed a deep learning model based on multi-scale dilated convolution feature fusion for the segmentation of black ice region in road image to send black ice warning to drivers in real time. In the proposed multi-scale dilated convolution feature fusion network, different dilated ratio convolutions are connected in parallel in the encoder blocks, and different dilated ratios are used in different resolution feature maps, and multi-layer feature information are fused together. The multi-scale dilated convolution feature fusion improves the performance by diversifying and expending the receptive field of the network and by preserving detailed space information and enhancing the effectiveness of diated convolutions. The performance of the proposed network model was gradually improved with the increase of the number of dilated convolution branch. The mIoU value of the proposed method is 96.46%, which was higher than the existing networks such as U-Net, FCN, PSPNet, ENet, LinkNet. The parameter was 1,858K, which was 6 times smaller than the existing LinkNet model. From the experimental results of Jetson Nano, the FPS of the proposed method was 3.63, which can realize segmentation of black ice field in real time.

• Tribology and Lubricants
• /
• v.36 no.1
• /
• pp.27-33
• /
• 2020

Self-equalizing tilting pad thrust bearings are usually employed in turbomachines to achieve high stability and reliability. A tilting pad bearing can incorporate self-equalizing links to handle the misalignment between the bearing and the thrust collar. In this popular design method, the pads sit on the upper-level plates and the lower-level plates stay on the retainer base. With misalignment, the pads that are heavily loaded are pushed down. Consequently, the link pushes up the pads on the opposite side, keeping the bearing surface parallel to the thrust collar surface. The self-equalizing link is used to handle the misalignment from the thermal and mechanical effects. In this study, the experimental investigation deals with the performance of self-equalizing tilting pad thrust bearings. The test rig for evaluating the performance of bearing is developed which can control the misalignment angle. Simultaneous measurements are taken for the force acting on each pad. Pad metal temperature and oil film thickness are functions of the shaft speed, bearing load, misalignment angle, and design of leveling plates. The effect of misalignment on bearing performance is discussed. The results demonstrate that the load on each pad depends on the test conditions(especially misalignment angle), and the load influences the performance of bearings.

• Journal of Internet Computing and Services
• /
• v.10 no.4
• /
• pp.199-211
• /
• 2009

Recent advancements in wireless networks have enabled support for mobile applications to transfer data over heterogeneous wireless paths in parallel using data striping technique [2]. Traditionally, high performance data transfer requires tuning of multiple TCP sockets, at sender's end, based on bandwidth delay product (BDP). Moreover, traditional techniques like Automatic TCP Buffer Tuning (ATBT), which balance memory and fulfill network demand, is designed for wired infrastructure assuming single flow on a single socket. Hence, in this paper we propose a buffer tuning technique at senders end designed to ensure high performance data transfer by striping data at transport layer across heterogeneous wireless paths. Our mechanism has the capability to become a resource management system for transport layer connections running on multi-homed mobile host supporting features for wireless link i.e. mobility, bandwidth fluctuations, link level losses. We show that our proposed mechanism performs better than ATBT, in efficiently utilizing memory and achieving aggregate throughput.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.3
• /
• pp.318-323
• /
• 2005

In this paper, we propose a systematic method of a fuzzy-model-based controller for continuous-time nonlinear dynamical systems which may contain uncertainties. The continuous-time uncertain TS fuzzy model is first constructed to represent the uncertain nonlinear system. A parallel distributed compensation (PDC) technique is then used to design a fuzzy model based controller for both stabilization and tracking. Finally, the designed continuous-time controller is converted to an equivalent discrete-time controller by using an intelligent digital redesign method. This new design technique provides a systematic and effective framework for integration of the fuzzy model based control theory and the advanced digital redesign technique for nonlinear dynamical systems with uncertainties. Finally, the single link flexible-joint robot arm is used as an illustrative example to show the effectiveness and the feasibility of the developed design method.

• The Transactions of the Korean Institute of Electrical Engineers P
• /
• v.63 no.4
• /
• pp.361-367
• /
• 2014

This paper is related to a new solar power conditioner for a built-in step-up chopper function. In the first step-up chopper proposed solar PV power conditioner for mutually connected in series with the input voltage of the bypass diodes are respectively connected to the positive terminal should install the mutual boosting chopper diode connected in series with the boost chopper switching element between the two power supply and at the same time the first and the second was connected to a second diode and a resonance inductor and a snubber capacitor in series with each other. And the common connection point between the bypass diode and the step-up chopper and the step-up chopper diode common connection point of the switching elements of the input voltage was set to the boost inductor for storing energy. In addition, between the step-up chopper and the step-up chopper diode and a switching element of a joint connection point of the first auxiliary diode and the second common connection point of the auxiliary diode was provided, the resonance capacitor. Between the step-up chopper and the step-up chopper diode and a switching element of a joint connection point and the common connection point of the resonance inductor snubber capacitor and connecting the third secondary diode, between two power supply lines is characterized by configuring the DC link capacitor bus lines in parallel. Therefore, it is possible to suppress the switching loss through, DC link bus lines, as well as there could seek miniaturization and weight reduction of the power conditioner itself by using a common capacitor of the non-polar non-polar electrolytic capacitor having a capacitor, the service life of the circuit can be extended and it is possible to greatly reduce the loss can be greatly improve the reliability of the product and the operation of the product itself.

• Journal of Electrical Engineering and Technology
• /
• v.2 no.1
• /
• pp.89-97
• /
• 2007

This paper presents a new circuit topology of active edge resonant snubbers assisted half-bridge soft switching PWM inverter type DC-DC high power converter for DC bus feeding power plants. The proposed DC-DC power converter is composed of a typical voltage source-fed half-bridge high frequency PWM inverter with a high frequency planar transformer link in addition to input DC busline side power semiconductor switching devices for PWM control scheme and parallel capacitive lossless snubbers. The operating principle of the new DC-DC converter treated here is described by using switching mode equivalent circuits, together with its unique features. All the active power switches in the half-bridge arms and input DC buslines can achieve ZCS turn-on and ZVS turn-off commutation transitions. The total turn-off switching losses of the power switches can be significantly reduced. As a result, a high switching frequency IGBTs can be actually selected in the frequency range of 60 kHz under the principle of soft switching. The performance evaluations of the experimental setup are illustrated practically. The effectiveness of this new converter topology is proved for such low voltage and large current DC-DC power supplies as DC bus feeding from a practical point of view.

• The KIPS Transactions:PartA
• /
• v.15A no.4
• /
• pp.189-198
• /
• 2008

In this paper, we prove mesh-like networks can be embedded into Petersen-Torus(PT) networks. Once interconnection network G is embedded in H, the parallel algorithm designed in Gcan be applied to interconnection network H. The torus is embedded into PT with dilation 5, link congestion 5 and expansion 1 using one-to-one embedding. The honeycomb mesh is embedded into PT with dilation 5, link congestion 2 and expansion 5/3 using one-to-one embedding. Additional, We derive average dilation. The embedding algorithm could be available in both wormhole routing system and store-and-forward routing system by embedding the generally known Torus and honeycomb mesh networks into PT at 5 or less of dilation and congestion, and the processor throughput could be minimized at simulation through one-to-one.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.8
• /
• pp.525-533
• /
• 2009

Staging technique is used to provide files for a job in the Grid. If a staged file has large volume, the start time of the job is delayed and the throughput of job in the Grid may decrease. Therefore, removal of staging overhead helps the Grid operate more efficiently. In this paper, we present two methods for efficient file provisioning to clear the overhead. First, we propose RA-RFT, which extends RFT of Globus Toolkit and enables it to utilize RLS with replica information. RA-RFT can reduce file transfer time by doing partial transfer for each replica in parallel. Second, we suggest Remote Link that uses remote I/O instead of file transfer. Remote link is able to save storage of computational nodes and enables fast file provisioning via prefetching. Through various experiments, we argue that our two methods have an advantage over existing staging techniques.

• Journal of KIISE:Computer Systems and Theory
• /
• v.29 no.2
• /
• pp.65-74
• /
• 2002

Recently star networks are considered as attractive alternatives to the widely used hypercube for interconnection networks in parallel processing systems by many researchers. One of the fundamental communication problems on star graph networks is broadcasing In this paper we consider the broadcasting problems in star graph networks using wormhole routing. In wormhole routed system minimizing link contention is more critical for the system performance than the distance between two communicating nodes. We use Hamiltonian paths in star graph to set up link-disjoint communication paths We present a broadcast algorithm in n-dimensional star graph of N(=n!) nodes such that the total completion time is no larger than $([long_n n!]+1)$ steps where $([long_n n!]+1)$ is the lower bound This result is significant improvement over the previous n-1 step broadcasting algorithm.