• Journal of Advanced Navigation Technology
• /
• v.24 no.5
• /
• pp.430-437
• /
• 2020

The proposed model implements a model that improves the face prediction rate and recognition rate through learning with an artificial neural network using face detection, landmark and face recognition algorithms. After landmarking in the face images of a specific person, the proposed model use the previously learned Caffe model to extract face detection and embedding vector 128D. The learning is learned by building machine learning algorithms such as support vector machine (SVM) and deep neural network (DNN). Face recognition is tested with a face image different from the learned figure using the learned model. As a result of the experiment, the result of learning with DNN rather than SVM showed better prediction rate and recognition rate. However, when the hidden layer of DNN is increased, the prediction rate increases but the recognition rate decreases. This is judged as overfitting caused by a small number of objects to be recognized. As a result of learning by adding a clear face image to the proposed model, it is confirmed that the result of high prediction rate and recognition rate can be obtained. This research will be able to obtain better recognition and prediction rates through effective deep learning establishment by utilizing more face image data.

• Journal of KIISE:Software and Applications
• /
• v.30 no.3_4
• /
• pp.251-258
• /
• 2003

A support vector machine (SVM) is supposed to provide a good generalization performance, but the actual performance of a actually implemented SVM is often far from the theoretically expected level. This is largely because the implementation is based on an approximated algorithm, due to the high complexity of time and space. To improve this limitation, we propose ensemble of SVMs by using Bagging (bootstrap aggregating) and Boosting. By a Bagging stage each individual SVM is trained independently using randomly chosen training samples via a bootstrap technique. By a Boosting stage an individual SVM is trained by choosing training samples according to their probability distribution. The probability distribution is updated by the error of independent classifiers, and the process is iterated. After the training stage, they are aggregated to make a collective decision in several ways, such ai majority voting, the LSE(least squares estimation) -based weighting, and double layer hierarchical combining. The simulation results for IRIS data classification, the hand-written digit recognition and Face detection show that the proposed SVM ensembles greatly outperforms a single SVM in terms of classification accuracy.

• Journal of Communications and Networks
• /
• v.10 no.3
• /
• pp.316-330
• /
• 2008

This paper proposes a novel method for transporting various types of user traffic effectively over the next generation network called integrated services digital network 3 (ISDN3) (or quantum network) using quantum packets. Basically, a quantum packet comprises one or more 53-byte quanta as generated by a "quantumization" process. While connection-oriented traffic is supported by fixed-size quantum packets each with one quantum to emulate circuit switching, connectionless traffic (e.g., IP packets and active packets) is carried by variable-size quantum packets with multiple quanta to support store-and-forward switching/routing. Our aim is to provide frame-like or datagram-like services while enabling cell-based multiplexing. The quantum packet method also establishes a flexible and extensible framework that caters for future packetization needs while maintaining backward compatibility with ATM. In this paper, we discuss the design of the quantum packet method, including its format, the "quantumization" process, and support for different types of user traffic. We also present an analytical model to evaluate the consumption of network resources (or network costs) when quantum packets are employed to transfer loss-sensitive data using three different approaches: cut-through, store-and-forward and ideal. Close form mathematical expressions are obtained for some situations. In particular, in terms of network cost, we discover two interesting equivalence phenomena for the cut-through and store-and-forward approaches under certain conditions and assumptions. Furthermore, analytical and simulation results are presented to study the system behavior. Our analysis provides valuable insights into the. design of the ISDN3/quantum network.

• Journal of the Korea Society for Simulation
• /
• v.15 no.3
• /
• pp.87-91
• /
• 2006

With the growth of the internet, the number of wireless internet users has increased continuously up to date. However, mobile communications could not support high speed transmission rate with cheap communication fee and wireless LAN has problems in providing terminal mobility and wide area connectivity, respectively. So the WMAN standard has been newly designed to make up for these limits. The initial 802.16 specification effectively offers a solution for providing fixed users with high speed wireless communication but it does not offer terminal mobility. So the 802.16d and 802.16e have been developed as the next generation solution that can support various PHY layer (SC, SCa, OFDM, OFDMA) and offer the terminal mobility. In this paper, we propose an effective uplink scheduling algorithm for VoIP with using UGS, and we show that our proposed algorithm is superior in view of average delay and management of uplink bandwidth to conventional rtPS algorithm and the scheme in reference, with using NS-2 network simulator.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.10 no.5
• /
• pp.812-820
• /
• 2006

Ethernet has been given a greater attention recently due to tendency of unifying most of transmission technique(not only LAN, but MAN and WAN) to ethernet. Performance evaluation was performed using C language for 10Gbps ethernet Data Link to design the optimum hardware, then internal FIFO size was evaluated. In this paper, MAC core for 10Gbps ethernet which contains high layer interface, transmit engine, flow control block, receive engine, reconciliation sublayer, configuration block, statistics block, and XGMII interface block was designed using VHDL language and Xilinx 6.2i tool and verified using Model_SIM 5.7G simulator. According to the specification of 10Gbps ethernet, MAC core with 64-bit data path should support 156.25MHz in order to support 10Gbps. The designed MAC core that process 64-bit data, operates at 168.549MHz and hence supports the maximum 10.78Gbps data processing. The designed MAC core is applicable to an area that needs a high-speed data processing of 10Gbps or more.

• Membrane Journal
• /
• v.20 no.3
• /
• pp.210-216
• /
• 2010

The hydrogen permselective membrane were prepared by chemical vapor deposition (CVD) aiming at the applications to hydrogen iodide decomposition in the thermochemical IS process, and it was evaluated for the possibility as a separation membrane. An electron probe X-ray microanalyzer (EPMA) and SEM picture were used to analyze the morphology and structure of the prepared membranes. It was confirmed that Zr-Si-O layer exist in the surface of the prepared membrane using zirconia coated support. Single-component permeance to $H_2$ and $N_2$ were measured at $300{\sim}600^{\circ}C$. Hydrogen permeance through the Z-1 membrane at a permeation temperature of $600^{\circ}C$ was about $1{\times}10^{-7}\;mol{\cdot}Pa^{-1}{\cdot}m^{-2}{\cdot}s^{-1}$. The selectivities of $H_2/N_2$ at $600^{\circ}C$ were 5.0 and 5.75 for Z-1 and Z-2 membrane, respectively.

• Korean Journal of Organic Agriculture
• /
• v.20 no.3
• /
• pp.267-282
• /
• 2012

In this study, RPS system, one of the renewable energy support systems, is utilized for economic analysis of solar generation equipment and the fuel cost savings plan for controlled horticulture farms with high fuel-cost dependency and facility applicability were evaluated. On the exterior of the upper layer of glass greenhouse (9917$m^2$) of controlled horticulture farms using bunker C oil, half of the area (4958$m^2$) was utilized for theoretical installation and operation of 450kW-level solar power generator, and as the result, first, the effect of investment cost only of solar generation system was found to be quite excellent, but it was analyzed that there were limits to saving the fuel costs of the controlled horticulture farms. Second, when geothermal system was first introduced in the farm and solar system was additionally introduced, it was analyzed that the effect of introducing solar system was excellent. In order to apply such effects to the sites of farming, partial supplementation of RPS system which is being uniformly applied regardless of the purpose of renewable energy is necessary. When the subject of use directly install facilities where it is directly connected to national added-value such as food security created by the farming industry, it is necessary to introduce appropriate system that corresponds to such. Moreover, it was studied that the quick development of demonstrative complex that can practically evaluate the applicability of renewable energy in farming industry and interest and preparation of related institutions in financial support structure for its site application would lead to success.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2011.08a
• /
• pp.217-217
• /
• 2011

Recently, demand for thermally stable metal nanoparticles suitable for chemical reactions at high temperatures has increased to the point to require a solution to nanoparticle coalescence. Thermal stability of metal nanoparticles can be achieved by adopting core-shell models and encapsulating supported metal nanoparticles with mesoporous oxides [1,2]. However, to understand the role of metal-support interactions on catalytic activity and for surface analysis of complex structures, we developed a novel catalyst design by coating an ultra-thin layer of titania on Pt supported silica ($SiO_2/Pt@TiO_2$). This structure provides higher metal dispersion (~52% Pt/silica), high thermal stability (~600$^{\circ}C$) and maximization of the interaction between Pt and titania. The high thermal stability of $SiO_2/Pt@TiO_2$ enabled the investigation of CO oxidation studies at high temperatures, including ignition behavior, which is otherwise not possible on bare Pt nanoparticles due to sintering [3]. It was found that this hybrid catalyst exhibited a lower activation energy for CO oxidation because of the metal-support interaction. The concept of an ultra-thin active metal oxide coating on supported nanoparticles opens-up new avenues for synthesis of various hybrid nanocatalysts with combinations of different metals and oxides to investigate important model reactions at high-temperatures and in industrial reactions.

• Membrane Journal
• /
• v.24 no.4
• /
• pp.292-300
• /
• 2014

Chlorine-resistant sulfonated poly(arylene ether sulfone) random copolymer (SPAES)-thin film composite (TFC) membranes for desalination are prepared using monoglyme as a selective solvent, which dissolves SPAES, but should be inert to porous polysulfone layer (e.g., Udel$^{(R)}$). Different from formic acid and diethylene glycol used as other selective solvents, monoglyme is environmentally friendly and has much lower boiling temperature. After a pretreatment of Udel$^{(R)}$ support film in isopropyl alcohol-glycerine mixture to minimize pore penetration leading to fairly reduced water flux, coating of SPAES solution in monoglyme onto the support and stepwise drying processes are conducted for defect-free TFC formation. The transport behavior through SPAES-TFC membranes is observed, correlating with the effects of sulfonation level, protonation, and physical and chemical crosslinking of SPAES selective layers.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.7
• /
• pp.275-280
• /
• 2018

As the global population becomes aging, the demand for health services for the elderly is expected to increase. In particular, The elderly visiting the emergency department sometimes have complex medical, social, and physical problems, such as having a variety of illnesses or complaints of unusual symptoms. The proposed system is designed to predict the mortality of the elderly patients who are over 65 years old and have admitted the emergency department. For mortality prediction, we compare the support vector machines and Feed Forward Neural Network (FFNN) trained with medical data such as age, sex, blood pressure, body temperature, etc. The results of the FFNN with a hidden layer are best in the mortality prediction, and F1 score and the AUC is 52.0%, 88.6% respectively. If we improve the performance of the proposed system by extracting better medical features, we will be able to provide better medical services through an effective and quick allocation of medical resources for the elderly patients visiting the emergency department.