• Annual Conference on Human and Language Technology
• /
• 2017.10a
• /
• pp.116-121
• /
• 2017

cQA(Community-based Question Answering) 시스템은 온라인 커뮤니티를 통해 사용자들이 질문을 남기고 답변을 작성할 수 있도록 만들어진 시스템이다. 신규 질문이 인입되면, 기존에 축적된 cQA 저장소에서 해당 질문과 가장 유사한 질문을 검색하고, 그 질문에 대한 답변을 신규 질문에 대한 답변으로 대체할 수 있다. 하지만, 키워드 매칭을 사용하는 전통적인 검색 방식으로는 문장에 내재된 의미들을 이용할 수 없다는 한계가 있다. 이를 극복하기 위해서는 의미적으로 동일한 문장들로 학습이 되어야 하지만, 이러한 데이터를 대량으로 확보하기에는 어려움이 있다. 본 논문에서는 질문이 제목과 내용으로 분리되어 있는 대량의 cQA 셋에서, 질문 제목과 내용을 의미 벡터 공간으로 사상하고 두 벡터의 상대적 거리가 가깝게 되도록 학습함으로써 의사(pseudo) 유사 의미의 성질을 내재화 하였다. 또한, 질문 제목과 내용의 의미 벡터 표현(representation)을 위하여, semi-training word embedding과 CNN(Convolutional Neural Network)을 이용한 딥러닝 기법을 제안하였다. 유사 질문 검색 실험 결과, 제안 모델을 이용한 검색이 키워드 매칭 기반 검색보다 좋은 성능을 보였다.

• Journal of Advanced Marine Engineering and Technology
• /
• v.32 no.8
• /
• pp.1263-1268
• /
• 2008

In this paper, the low noise power amplifier for GaAs FET ATF-10136 is designed and fabricated with active bias circuit and self bias circuit. To supply most suitable voltage and current, active bias circuit is designed. Active biasing offers the advantage that variations in the pinch-off voltage($V_p$) and saturated drain current($I_{DSS}$) will not necessitate a change in either the source or drain resistor value for a given bias condition. The active bias network automatically sets a gate-source voltage($V_{gs}$) for the desired drain voltage and drain current. Using resistive decoupling circuits, a signal at low frequency is dissipated by a resistor. This design method increases the stability of the LNA, suitable for input stage matching and gate source bias. The LNA is fabricated on FR-4 substrate with active and self bias circuit, and integrated in aluminum housing. As a results, the characteristics of the active and self bias circuit LNA implemented more than 13 dB and 14 dB in gain, lower than 1 dB and 1.1 dB in noise figure, 1.7 and 1.8 input VSWR at normalized frequency $1.4{\sim}1.6$, respectively.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.17 no.6 s.109
• /
• pp.511-518
• /
• 2006

In this paper, a small RFID tag antenna in UHF band which has bandwidth-enhanced characteristic is proposed. The shape of the proposed antenna is a meander antenna to have size-reduced characteristic, and it consists of two radiators which make dual resonance in adjacent frequency to enhance bandwidth. By adjusting length and location of each radiator, the proposed antenna can make dual resonance at arbitrary location on the Smith chart, which is able to make impedance matching with RFID tag chip in wide frequency range. And it is apparent that the proposed antenna can have bandwidth-enhanced characteristic according to the simulated and measured results.

• ETRI Journal
• /
• v.36 no.2
• /
• pp.214-223
• /
• 2014

A fully-integrated penta-band reconfigurable power amplifier (PA) is developed for handset Tx applications. The output structure of the proposed PA is composed of the fixed output matching network, power and frequency reconfigurable networks, and post-PA distribution switches. In this work, a new reconfiguration technique is proposed for a specific band requiring power and frequency reconfiguration simultaneously. The design parameters for the proposed reconfiguration are newly derived and applied to the PA. To reduce the module size, the switches of reconfigurable output networks and post-PA switches are integrated into a single IC using a $0.18{\mu}m$ silicon-on-insulator CMOS process, and a compact size of $5mm{\times}5mm$ is thus achieved. The fabricated W-CDMA PA module shows adjacent channel leakage ratios better than -39 dBc up to the rated linear power and power-added efficiencies of higher than around 38% at the maximum linear output power over all the bands. Efficiency degradation is limited to 2.5% to 3% compared to the single-band reference PA.

• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.6 no.1
• /
• pp.7-17
• /
• 1997

During the past decade, there were many well-established theories for the adaptive control of linear systems, but there exists relatively little general theory for the adaptive control of nonlinear systems. Adaptive control technique is essential for providing a stable and robust performance for application of industrial robot control. Neural network computing methods provide one approach to the development of adaptive and learning behavior in robotic system for manufacturing. Computational neural networks have been demonstrated which exhibit capabilities for supervised learning, matching, and generalization for problems on an experimental scale. Supervised learning could improve the efficiency of training and development of robotic systems. In this paper, a new scheme of adaptive-neuro control system to implement real-time control of robot manipulator using digital signal processors is proposed. Digital signal processors, DSPs, are micro-processors that are developed particularly for fast numerical computations involving sums and products of variables. The proposed neuro control algorithm is one of learning a model based error back-propagation scheme using Lyapunov stability analysis method. The proposed adaptive-neuro control scheme is illustrated to be an efficient control scheme for implementation of real-time control for SCARA robot with four-axes by experiment.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.12
• /
• pp.1244-1248
• /
• 2010

The previous study of Digital Broadcasting Recommendation system is based on user explicit profiling information. But user profile is always changing and the exact extraction of user profile is very important in recommendation system like Digital TV using many user interactions. This paper is studied of realtime user profiles aggregation through user remote controller input and matching this profiles with contents meta-data like contents genre information, event information, content viewing time. It is not used commercial database system and network communication solution considering embedded system hardware restriction. And it is considered people want different content genre based on watching time. From the results of this paper, there are improvement of user satisfaction of contents recommendation.

• Journal of Electrical Engineering and Technology
• /
• v.13 no.6
• /
• pp.2434-2440
• /
• 2018

A new type of piezoelectric micromachined ultrasonic transducer (pMUT) with high resonant frequency was developed by using a thin lead zirconate titanate (PZT) as an insulation layer on a floating $10{\mu}m$ silicon membrane. The PZT insulation layer facilitated acoustic impedance matching at active pMUT, leading to a high performance in the acoustic conversion property compared with the transducer using $SiO_2$ insulation layer. The fabricated ultrasonic devices were wirelessly measured by connecting two identical acoustic transducers to two separate ports in a single network analyzer simultaneously. The acoustic wave emitted from a transducer induced a $3.16{\mu}W$ on the other side of the transducer at a distance of 2 cm. The transducer performances in terms of device diameters, PZT thickness, annealings, and different DC polings, etc. were investigated. COMSOL simulation was also performed to predict the device performances prior to fabrication. Based on the COMSOL simulation, the device was fabricated and the results were compared.

• KIPS Transactions on Software and Data Engineering
• /
• v.6 no.6
• /
• pp.285-292
• /
• 2017

There are technological, operational and environmental constraints at tactical edge, which are disconnected operation, intermittent connectivity, and limited bandwidth (DIL), size, weight and power (SWaP) limitations, ad-hoc and mobile network, and so on. To overcome these limitations and constraints, we use service-oriented architecture (SOA) based technologies. In our research, we propose a hierarchical mission service model that supports service-oriented mission planning and execution in order for a commander to operate various SW required for mission in battlefield environment. We will also implement development tools that utilize the workflow technology and semantic capability-based recommendation and apply them to combat mission scenarios to demonstrate effectiveness.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1887-1891
• /
• 2005

Despite significant advances in multimedia transferring technologies in various fields of robotics, it is sometimes quite difficult for the operator to fully understand the context of 3D remote environments from 2D image feedback. Particularly, in the remote control of mobile robots, the recognition of the object associated with the task is very important, because the operator has to control the robot safely in various situations not through trial and error. Therefore, it is necessary to provide the operator with 3D volumetric models of the object and object-related information as well such as locations, shape, size, material properties, and so on. Thus, in this paper, we propose a vision-based human interface system that provides an interactive, information-rich map through network-based information brokering. The system consists of an object recognition part, a 3D map building part, a networked knowledge base part, and a control part of the mobile robot.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1287-1292
• /
• 2005

This paper presents an effective pattern classification model by designing an artificial neural network based pattern classifiers for face recognition. First, a RGB image inputted from a frame grabber is converted into a HSV image which is similar to the human beings' vision system. Then, the coarse facial region is extracted using the hue(H) and saturation(S) components except intensity(V) component which is sensitive to the environmental illumination. Next, the fine facial region extraction process is performed by matching with the edge and gray based templates. To make a light-invariant and qualified facial image, histogram equalization and intensity compensation processing using illumination plane are performed. The finally extracted and enhanced facial images are used for training the pattern classification models. The proposed H-ART2 model which has the hierarchical ART2 layers and F-LVQ model which is optimized by fuzzy membership make it possible to classify facial patterns by optimizing relations of clusters and searching clustered reference patterns effectively. Experimental results show that the proposed face recognition system is as good as the SVM model which is famous for face recognition field in recognition rate and even better in classification speed. Moreover high recognition rate could be acquired by combining the proposed neural classification models.