• Journal of KIISE
• /
• v.43 no.7
• /
• pp.795-800
• /
• 2016

In image classification, recent CNNs compete with human performance. However, there are limitations in more general recognition. Herein we deal with indoor images that contain too much information to be directly processed and require information reduction before recognition. To reduce the amount of data processing, typically variational inference or variational Bayesian methods are suggested for object detection. However, these methods suffer from the difficulty of marginalizing over the given space. In this study, we propose an image-text integrated recognition system using active vision based on Spatial Transformer Networks. The system attempts to efficiently sample a partial region of a given image for a given language information. Our experimental results demonstrate a significant improvement over traditional approaches. We also discuss the results of qualitative analysis of sampled images, model characteristics, and its limitations.

• Management & Information Systems Review
• /
• v.28 no.1
• /
• pp.125-156
• /
• 2009

The primary purpose of this study is to identify the determinants of the users' satisfaction and commitment in personal media. For these purposes, we developed a research model based on the literature reviews of personal media, perceived risk and interactivity, satisfaction, and commitment. This study has identified four dimensions in the concept of perceived risk, such as privacy risk, social risk, time loss risk, and economic risk, and three dimensions in the concept of perceived interactivity, such as active control, two-way communication, and responsiveness. In survey, data were collected from 428 male and female who have an experience in personal media. Data was analyzed factor analysis, reliability analysis, measurement model analysis, and structural model analysis by utilizing SPSS 15.0 and AMOS 5.0 program. Analysis results of research model indicate the following. Firstly, privacy risk, active control, two-way communication, and responsiveness are significantly related satisfaction. Secondly, privacy risk, time loss risk, economic risk, active control, and two-way communication are significantly related commitment. Thirdly, users' satisfaction has positive relationship with commitment in personal media. Based on these results, theoretical implications for relevant researchers and managerial implications for personal media vitalization and marketing strategy are discussed. Finally, limitations for this study and further research issues are suggested.

• Journal of KIISE:Software and Applications
• /
• v.30 no.5_6
• /
• pp.570-578
• /
• 2003

Maximum entropy models are promising candidates for natural language modeling. However, there are two major hurdles in applying maximum entropy models to real-life language problems, such as prepositional phrase attachment: feature selection and high computational complexity. In this paper, we propose a maximum entropy boosting model to overcome these limitations and the problem of imbalanced data in natural language resources, and apply it to prepositional phrase (PP) attachment and part-of-speech (POS) tagging. According to the experimental results on Wall Street Journal corpus, the model shows 84.3% of accuracy for PP attachment and 96.78% of accuracy for POS tagging that are close to the state-of-the-art performance of these tasks only with small efforts of modeling.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.37 no.2
• /
• pp.139-146
• /
• 2009

In this study, an assessment is made for the helicopter vibration reduction of composite rotor blades using an active twist control concept. The piezoceramic shear actuation mechanism along with elastic couplings of composite blades is used for vibration reduction. The rotor blades are modeled as composite box-beams with actuator layers bonded on the outer surfaces of the thin-walled section. The governing equations of motion for helicopter blades are obtained using Hamilton's principle. A time domain unsteady aerodynamic theory with free wake model is used to obtain the airloads. Various rotor configurations with different elastic couplings with appropriate actuator placement are used to investigate the hub vibration characteristics. Numerical results show that a substantial reduction of $N_b$/rev hub vibration can be achieved using the optimal control algorithm.

• Journal of Korean Library and Information Science Society
• /
• v.39 no.2
• /
• pp.291-312
• /
• 2008

This study intends to make a comprehensive investigation on an application of FRBR model to KORMARC records. To the end, issues associated with FRBR implementation are analyzed, and automatic algorithms to covert bibliographic records into FRBR model are examined thoroughly. On the basis of it. this study implemented four-level hierarchical bibliographic structure of FRBR model using existing KORMARC records. 161 sample records for an experiment were selected from Korea national library. The work chosen as sample records was "Hamlet." Results suggest that considerable difficulties would be encountered in the process of FRBRizing KORMARC records.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.18 no.5
• /
• pp.171-177
• /
• 2018

Recently there are many image datasets which has variety of data class and point to extract general features. But in order to this variety data class and point, deep learning model trained this dataset has not good performance in heterogeneous data feature local area. In this paper, we propose the structure which use sub-category and openset object detection methods to train more robust model, named multi-branch tree using ASSL. By using this structure, we can have more robust object detection deep learning model in heterogeneous data feature environment.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.99-108
• /
• 2021

In this paper numerical wave propagation experiments have been performed to visually confirm whether the signals scattered by water masses can be a false alarm in active sonar. The numerical environments consist of exaggerated water masses as targets in free space. Using a pseudospectral time-domain model for irregular boundary, the back-scattered signals have been calculated and compared with analytic solutions. Also, the sound propagation was simulated. Consequently, it was verified that water masses themselves could not be detected as a false target.

• Proceedings of the Korean Society of Applied Pharmacology
• /
• 1994.04a
• /
• pp.291-291
• /
• 1994

Ibotonic acid와 FeCl$_3$ 투여군은 phosphate buffered saline을 투여한 대조군에 비해 흰쥐의 능동적 회피학습 습득율을 감소시켰고 해마영역의 과산화지질 형성도를 유의하게 증가시켰으며 acetylcholinesterase의 활성을 감소시켰으나 acetylcholine의 농도에는 별영향을 주지 못하였다. 이들 약물은 신경세포 지질막등에 작용하여 노인성 치매에서와 유사한 cholinergic신경계통에 neuronal degeneration을 유발하여 학습능력 감소를 유발한 것으로 보인다. 4,결론 중추 Cholinergic성 신경계활성의 변화와 뇌내 과산화지질의 과다 형성이 학습능력 감소 및 노인성치매의 원인이라고 말할수는 없지만 이러한 뇌내의 변화률이 치매과정과 연관성을 갖는 것으로 보여진다.

• Journal of KSNVE
• /
• v.11 no.2
• /
• pp.292-300
• /
• 2001

This research is concerned with the validation of the modeling technique and controller design for slewing beam structures. When cantilever beam rotates about axes perpendicular to the undeformed beam's longitudinal axis, it experiences inertial loading. Hence, the beam vibrates from the initial stage of slewing. In this paper, the analytical model for a single slewing flexible beam with surface bonded piezoelectric sensor and actuator is developed using the Hamilton's principle with discretization by the assumed mode method. Comparisons with the theoretical model are made based upon the frequency responses and time responses. A new factor called the coupling coefficient is introduced to incorporate the discrepancies between the theoretical and experimental results. The slewing is achieved by applying the PID control, which is found to be less sensitive to vibrations. The vibrations are controlled by PPF controller, which is found to be effective in suppressing residual vibrations after slewing. The vibrations occurred during slewing is difficult to control because the piezoceramic actuator is not powerful enough to overcome inertial loadings.

• Journal of the Earthquake Engineering Society of Korea
• /
• v.10 no.6 s.52
• /
• pp.93-102
• /
• 2006

Control input's saturation of active control devices for large structures under large external disturbances are often occurred. It is more difficult to obtain the exact values of mass and stiffness as structures are higher. The modelling errors between mathematical models and real structures must be also included as parameter uncertainties. Therefore, in active vibration control of civil engineering structures like buildings and bridges, the robust saturation controller design method considering both control input's saturation and parameter uncertainties of system is needed. In this paper, stabilities of linear optimal controller LQR, modified bang-bang controller, saturated sliding mode controller, and robust saturation controller among various controllers which have been studied and applied to active vibration control of buildings are investigated. Especially, unstable phenomena of the LQR, the modified bang-bang controller and the saturated sliding mode controller when the control input is saturated or parameter uncertainties exist are presented to show the necessity of the robust saturation controller. The robust stability of the robust saturation controller are shown through a numerical example of a 2DOF linear vibrating system and an experimental test of the two-story structure with an active mass damper (AMD).