• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.320-323
• /
• 2022

상호참조해결은 문서에서 명사, 대명사, 명사구 등의 멘션 후보를 식별하고 동일한 개체를 의미하는 멘션들을 찾아 그룹화하는 태스크이다. 딥러닝 기반의 한국어 상호참조해결 연구들에서는 BERT를 이용하여 단어의 문맥 표현을 얻은 후 멘션 탐지와 상호참조해결을 동시에 수행하는 End-to-End 모델이 주로 연구가 되었으며, 최근에는 스팬 표현을 사용하지 않고 시작과 끝 표현식을 통해 상호참조해결을 빠르게 수행하는 Start-to-End 방식의 한국어 상호참조해결 모델이 연구되었다. 최근에 한국어 상호참조해결을 위해 구축된 ETRI 데이터셋은 WIKI, QA, CONVERSATION 등 다양한 도메인으로 이루어져 있으며, 신규 도메인의 데이터가 추가될 경우 신규 데이터가 추가된 전체 학습데이터로 모델을 다시 학습해야 하며, 이때 많은 시간이 걸리는 문제가 있다. 본 논문에서는 이러한 상호참조해결 모델의 도메인 적응에 Continual learning을 적용해 각기 다른 도메인의 데이터로 모델을 학습 시킬 때 이전에 학습했던 정보를 망각하는 Catastrophic forgetting 현상을 억제할 수 있음을 보인다. 또한, Continual learning의 성능 향상을 위해 2가지 Transfer Techniques을 함께 적용한 실험을 진행한다. 실험 결과, 본 논문에서 제안한 모델이 베이스라인 모델보다 개발 셋에서 3.6%p, 테스트 셋에서 2.1%p의 성능 향상을 보였다.

• Convergence Security Journal
• /
• v.22 no.1
• /
• pp.99-111
• /
• 2022

With current battlefield environment relying heavily on Network Centric Warfare(NCW), existing weaponary systems are evolving into a new concept that converges IT technology. Majority of the weaponary systems are implemented with numerous embedded softwares which makes such softwares a key factor influencing the performance of such systems. Furthermore, due to the advancements in both IoT technoogies and embedded softwares cyber threats are targeting various embedded systems as their scope of application expands in the real world. Weaponary systems have been developed in various forms from single systems to interlocking networks. hence, system level cyber security is more favorable compared to application level cyber security. In this paper, a secure real-time operating system has been designed, implemented and measured to protect embedded softwares used in weaponary systems from unknown cyber threats at the operating system level.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2005.11a
• /
• pp.331-334
• /
• 2005

현재 무선 이동통신의 발달로 고속 신뢰성 높은 데이터 전송을 요구하고 있다. 그러나 무선 이동통신 환경은 데이터 전송 시 지연과 간섭 등에 의해 주파수 선택성 페이딩을 가진다. OFDM은 주파수 선택성 페이딩에 영향을 받는 통신 시스템에서 채용되는 강력한 기술이다. OFDM에 적응 변조와 함께 송수신기에 다중의 안테나를 설치함으로서 채널 지연 확산에 강력히 대응한다. 본 논문의 연구는 MIMO 시스템에 적용된 적응 변조를 가진 OFDM이다. OFDM 각 서브채널의 상태에 따라 최적의 비트값을 할당하고 전력을 제어한다. 본 논문에서 제안한 적응비트로딩 MIMO-OFDM 시스템을 사용하면 현재의 시스템보다 더 좋은 BER을 가지며 고속 통신할 수 있다.

• Proceedings of the KSR Conference
• /
• 2011.10a
• /
• pp.1664-1666
• /
• 2011

오늘날 고성능의 소형 센서 및 무선통신 기술의 발달로 유비쿼터스 컴퓨팅의 실현이 가능하게 되었다. 미래의 스마트한 디바이스뿐만 아니라 무선통신이 가능한 USN(Ubiquito us Sensor network) 기술은 주변 현황을 인식하고 필요한 정보를 처리하여 현장건설 등에 피이드백 시킴으로써 보다 나은 건설 진행과정에 관한 파악과 설계변경 및 계획 등에 필요한 정보를 제공할 수 있다. 본 연구는 TinyOS 기반에서 운용되는 무선 통신에 의한 USN 기술과 그래픽 기반의 LabView 프로그래밍 기술을 융합하여 정보를 처리할 수 있는 일련의 인터페이스 방법을 구현하였다. 송수신된 데이터 처리 결과는 TinyOS 기반으로 동작하는 PC에 그래프 등으로 나타나도록 하였으며, 무선통신용 USN 기술과 융합된 그래픽 처리 기반의 마이크로프로세서 시스템의 장점과 편리성으로 건설현장의 진행과정파악 및 변경 등에 필요한 정보를 제공하며 건설현장 정보의 피이드백을 가능하도록 하였다. 그 결과, 철도건설 현장관리에서의 USN과 구조물의 정밀진단 및 관리에 매우 유용함을 입증하였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.3B
• /
• pp.398-407
• /
• 2010

Due to increase of user requirement for various traffics and the advance of network technology, each distinct network has converge into FMC(Fixed Mobile Convergence) networks. However, we need to research the performance analysis of VoIP(Voice over Internet Protocol) in the FMC network to provide QoE for the voice user of FMC network. Therefore, this paper introduces the scenario which is the situation of voice quality degradation when a user uses VoIP to communicate with other users in the FMC network. Especially, this paper presents scenario in terms of the component of the network and finds the improvement point of voice quality. In the simulation results, three improvement points of voice quality are found as following: voice quality degradation by packet loss in the physical layer of the HSDPA network, by utilizing GGSN without QoS parameter mapping mechanism which is gateway between 3GPP and IP backbone, and by using non-QoS AP in the WLAN network.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.6
• /
• pp.165-171
• /
• 2022

Human heart rate can be used to measure exercise intensity as an important indicator. If heart rate can be predicted, exercise can be performed more efficiently by regulating the intensity of exercise in advance. In this paper, a FitRec-based prediction model is proposed for estimating running heart rate for users. Endomondo data is utilized for training the proposed prediction model. The processing algorithms for time-series data, such as LSTM(long short term memory) and GRU(gated recurrent unit), are employed to compare their performance. On the basis of simulation results, it was demonstrated that the proposed model trained with running exercise performed better than the model trained with several cardiac exercises.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.1
• /
• pp.301-307
• /
• 2024

This study introduces a novel cascade fusion architecture aimed at enhancing thermal images across various scale conditions. The processing of thermal images at multiple scales has been challenging due to the limitations of existing methods that are designed for specific scales. To overcome these limitations, this paper proposes a unified framework that utilizes cascade feature fusion to effectively learn multi-scale representations. Confidence maps from different image scales are fused in a cascaded manner, enabling scale-invariant learning. The architecture comprises end-to-end trained convolutional neural networks to enhance image quality by reinforcing mutual scale dependencies. Experimental results indicate that the proposed technique outperforms existing methods in multi-scale thermal image enhancement. Performance evaluation results are provided, demonstrating consistent improvements in image quality metrics. The cascade fusion design facilitates robust generalization across scales and efficient learning of cross-scale representations.

• Journal of the Institute of Convergence Signal Processing
• /
• v.2 no.2
• /
• pp.62-67
• /
• 2001

In the recent mobile communication systems, the performance of Turbo Code using the error correction coding depends on the interleaver influencing the free distance determination and the recursive decoding algorithms that is executed in the turbo decoder. However, performance depends on the interleaver depth that need a large time delay over the reception process. Moreover, Turbo Code has been known as the robust ending method with the confidence over the fading channel. The International Telecommunication Union(ITU) has recently adopted as the standardization of the channel coding over the third generation mobile communications such as IMT-2000. Therefore, in this paper, we proposed of the method to improve the conventional performance with the parallel concatenated 4-New Turbo Decoder using MAP a1gorithm in spite of complexity increasement. In the real-time video and video service over the third generation mobile communications, the performance of the proposed method was analyzed by the reduced decoding delay using the variable decoding method by computer simulation over AWGN and fading channels.

• Journal of Convergence for Information Technology
• /
• v.10 no.8
• /
• pp.7-14
• /
• 2020

In the fifth generation (5G) mobile networks, the mobile services require 100 times faster connections. One of the promising 5G technologies is non-orthogonal multiple access (NOMA). In NOMA, the users share the channel resources, so that the more users can be served simultaneously. There are several advantages offered by NOMA, such as higher spectrum efficiency and low transmission latency, compared to orthogonal multiple access (OMA), which is usually used in the fourth generation (4G) mobile networks, for example, long term evolution (LTE). In this paper, we compare the receivers for NOMA. The standard NOMA receiver, the non-SIC NOMA receiver, and the symmetric superposition coding (SC) NOMA receiver are compared. Specifically, it is shown that the performance of the standard receiver is the best, whereas the performances of the non-SIC receiver and symmetric SC receiver are dependent on the power allocation.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.8 no.4
• /
• pp.75-88
• /
• 2008

This paper presents a spatial-level nonparametric multi-focus image fusion technique based on kernel estimates of input image blocks' underlying class-conditional probability density functions. Image fusion is approached as a classification task whose posterior class probabilities, P($wi{\mid}Bikl$), are calculated with likelihood density functions that are estimated from the training patterns. For each of the C input images Ii, the proposed method defines i classes wi and forms the fused image Z(k,l) from a decision map represented by a set of $P{\times}Q$ blocks Bikl whose features maximize the discriminant function based on the Bayesian decision principle. Performance of the proposed technique is evaluated in terms of RMSE and Mutual Information (MI) as the output quality measures. The width of the kernel functions, ${\sigma}$, were made to vary, and different kernels and block sizes were applied in performance evaluation. The proposed scheme is tested with C=2 and C=3 input images and results exhibited good performance.