DUALITY OF QK-TYPE SPACES

Mujun Zhan,Guangfu Cao 대한수학회 2014 대한수학회보 Vol.51 No.5

For BMO, it is well known that VMO∗∗ = BMO. In this paper such duality results of QK-type spaces are obtained which generalize the results by M. Pavlovi´c and J. Xiao.

A Relay Selection and Power Allocation Scheme for Cooperative Wireless Sensor Networks

( Mujun Qian ),( Chen Liu ),( Youhua Fu ),( Weiping Zhu ) 한국인터넷정보학회 2014 KSII Transactions on Internet and Information Syst Vol.8 No.4

This paper investigates optimal relay selection and power allocation under an aggregate power constraint for cooperative wireless sensor networks assisted by amplify-and-forward relay nodes. By considering both transmission power and circuit power consumptions, the received signal-to-noise ratio (SNR) at the destination node is calculated, based on which, a relay selection and power allocation scheme is developed. The core idea is to adaptively adjust the selected relays and their transmission power to maximize the received SNR according to the channel state information. The proposed scheme is derived by recasting the optimization problem into a three-layered problem ― determining the number of relays to be activated, selecting the active relays, and performing power allocation among the selected relays. Monte Carlo simulation results demonstrate that the proposed scheme provides a higher received SNR and a lower bit error rate as compared to the average power allocation scheme.

악성코드의 이미지화 방법이 딥러닝 기반의 악성코드 분류 성능에 미치는 영향 분석

송무준(Mujun Song),이종관(Jongkwan Lee) 육군사관학교 화랑대연구소 2021 한국군사학논집 Vol.77 No.1

In this paper, we present the effect of the interpolation method and image size, which are an essential process for pre-processing in deep learning, on the performance of deep learning-based malware classification. The classifier uses a deep learning model to classify malware, and the visualized malware samples are used as train and test datasets. The malware images should be the same size to be used as input data in the deep learning model. Then, the malware size is not the same, so that resizing of images is necessary. When the image is resized, the image features can be distorted depending on the interpolation methods and the target size of the image. We conduct experiments to understand the impact on the deep learning model’s classification performance under various conditions. The results of the experiments can guide for selecting interpolation method and image size for the deep learning-based malware classifier.

DUALITY OF Q_K-TYPE SPACES

Zhan, Mujun,Cao, Guangfu Korean Mathematical Society 2014 대한수학회보 Vol.51 No.5

For BMO, it is well known that $VMO^{**}=BMO$. In this paper such duality results of $Q_K$-type spaces are obtained which generalize the results by M. Pavlovi$\acute{c}$ and J. Xiao.

개인전투체계의 표적 식별 능력 향상을 위한 딥러닝 모델 연구

이종관(Jongkwan Lee),송무준(Mujun Song),김용철(Yongchul Kim),박상준(Sangjun Park),김도경(Do-Kyoung Kim),백승호(Seung-Ho Baek) 한국정보통신학회 2020 한국정보통신학회 종합학술대회 논문집 Vol.24 No.2

Specific disruption of contextual memory recall by sparse additional activity in the dentate gyrus

Cho, Hye-Yeon,Kim, Mujun,Han, Jin-Hee Elsevier 2017 INC. Vol.145 No.-

<P><B>Abstract</B></P> <P>The dentate gyrus (DG) of the hippocampus is essential for contextual and spatial memory processing. While lesion or silencing of the DG impairs contextual memory encoding and recall, overly activated DG also prevents proper memory retrieval. Abnormally elevated activity in the DG is repeatedly reported in amnesic mild cognitive impairment (aMCI) patients or aged adults. Although the correlation between memory failure and abnormally active hippocampus is clear, their causal relationship or the underlying nature of such interfering activity is not well understood. Using optogenetics aided by a carefully controlled adeno-associated virus infection system, we were able to examine the differential effects of abnormally activated hippocampus on mice motor behavior and memory function, depending on the extent of the stimulation. Optogenetic stimulation of massive proportion of dorsal DG cells resulted in memory retrieval impairment, but also induced increase in general locomotion. Random additional activity in a sparse population of dorsal DG neurons, however, interfered with contextual memory recall without inducing hyperactivity. Our findings thus establish the causal role of elevated DG activity on memory recall failure, suggesting such aberrant DG activity may contribute to amnesic symptoms in aMCI patients and aged adults.</P> <P><B>Highlights</B></P> <P> <UL> <LI> Massive activation of the dentate gyrus (DG) induces hyperactive behavior in mice. </LI> <LI> Sparse DG activation specifically impairs contextual fear memory recall. </LI> <LI> The impairment in memory retrieval is due to “additional” aberrant activity in DG. </LI> <LI> Such findings parallel the symptoms found in aged adults and Alzheimer’s patients. </LI> </UL> </P>

BCI2000기반 근전도 및 동작 상상 뇌전도를 이용한 무선 로봇 제어

조호현(Hohyun Cho),정찬민(Chanmin Jung),김무준(Mujun Kim),안민규,전성찬(Sung Chan Jun) 한국정보과학회 2011 한국정보과학회 학술발표논문집 Vol.38 No.2B

뇌-컴퓨터 인터페이스 기술은 인간 신경계의 중추인 두뇌와 컴퓨터 간의 직접적인 교류를 위한 기술로 이를 응용하면 신체적 장애를 가지고 있는 사람들에게 신체적 한계를 극복하게 해줄 수 있는 여러 보조 장치들을 개발할 수 있다. 이 연구는 생체 신호를 이용한 로봇 제어 연구로 Lego NXT 임베디드 시스템을 통하여 하드웨어 구현을 하였으며 기본적으로 이동을 위한 동작 외에 사용자가 뇌파로 제어하기에 정교함이 요구되는 기능들을 내장시키고 전체적인 제어는 MFC MatLab 그리고 뇌-컴퓨터 인터페이스 연구의 범용 플랫폼인 BCI2000 시스템을 사용하였다. 본 연구는 뇌파 및 근전도 신호를 통해 Lego NXT를 제어할 수 있는 BCI2000 기반 어플리케이션 제작에 관한 것으로 특별히 뇌전도 신호 처리 과정에서 3 클래스 FLDA(Fisher Linear Discriminant Analysis) 분석 기법을 도입하여 제어 안정화와 기능 확장을 할 수 있었다. 이번 연구 단계에서는 Lego NXT제어를 위한 전체적인 시스템 구축을 하였으며 차후 보다 효과적인 명령패러다임과 신호처리 방법을 도입한다면 뇌-컴퓨터 인터페이스 기반의 컨텐츠로 확장을 할 수 있을 것으로 기대가 된다.

Synthesis of 2,5-Furandicarboxylic Acid-Based Heat-Resistant Polyamides Under Existing Industrialization Process

Min Cao,Chuanhui Zhang,Bo He,Mujun Huang,Sujun Jiang 한국고분자학회 2017 Macromolecular Research Vol.25 No.7

As a bio-based material, 2,5-furandicarboxylic acid (FDCA) has been used widely as alternative material to terephthalic acid to produce polyesters or polyamides. In this study, PA10T/10F with melting point higher than 280 oC was produced under existing industrialization process (prepolymerization+solid state polymerization (SSP)), viscosity of which was affected greately by decarboxylation of FDCA. TGIR technique was used to study decarboxylation, and 1,10-decanediamine, CO2, H2O and furan derivates can be found, verifying decarboxylation scheme. 1H NMR, 13C NMR and 2D NMR analysis showed that nearly half of the feeding FDCA monomers decarboxylates during SSP, instead of entering the polyamide chain, leading to the lower furan moiety ratio in polymers compared to that of feeding ratio. However, thermal stability of PA10T/10F polyamide is comparable to PA10T homopolymer, which indicates that FDCA can be a promising base material for heat-resistant polyamide production.

Numerical Study of Temperature Distribution Control in Precision Glass Molding Furnace

Yang Hu,Lianguan Shen,Jian Zhou,Mujun Li 한국정밀공학회 2018 International Journal of Precision Engineering and Vol.19 No.6

In a precision glass molding process, glass preform is compressed at a high temperature well above its transition temperature and the temperature distribution inside plays an important role in determining the quality of final products. In this research, a 2D axisymmetric numerical heat transfer model, integrated with an innovative PID control subroutine, was employed for simulating the temperature control process and obtaining transient temperature distribution in the glass preform. Feasibility of this method was validated by experiments with good agreements. Finally, influence of air gap between glass preform and upper mold as well as temperature control mode were investigated. The results showed that temperature difference of the molds had a severer influence on the temperature distribution of glass preform with the decrease of air gap. In addition, a two-point control strategy was demonstrated as a valid method to improve the temperature uniformity in the glass preform.

뇌-컴퓨터 인터페이스 시스템에서 피험자의 잠재적 성능 확인을 위한 상상 운동움직임의 정확도 분석

안민규(Minkyu Ahn),조호현(Hohyun Cho),안상태(Sangtae Ahn),김무준(Mujun Kim),전성찬(Sung Chan Jun) 한국HCI학회 2012 한국HCI학회 학술대회 Vol.2012 No.1

뇌-컴퓨터 인터페이스를 위해 상상 손 움직임 명령 패러다임은 많이 사용되는 방법 중의 하나이다. 2클래스 명령을 위하여 주로 많이 사용되며 왼손/오른손 상상움직임을 통해 다르게 나타나는 뇌파의 공간적 패턴을 특징 값으로 사용한다. 하지만 손 상상 움직임에 대한 것은 피험자에 따라 특징이 잘 나타나지 않는 경우도 있기 때문에, 피험자의 잠재적 성능을 가늠해 보는 것은 중요하다고 할 수 있다. 이러한 피험자의 잠재적 가능성을 알아보기 위하여 52건의 왼손/오른손 상상실험으로부터 얻은 데이터를 분석하여, 다양한 조건에 대한 성능 비교 분석을 하였다. 결과적으로 대부분의 피험자가 무작위 레벨인 60% 부근에 분포함을 보였고, 100% 로 갈수록 피험자의 분포 값이 줄어듦을 확인 하였다. 조건에 대한 성능은 여성이 남성에 비하여 좋은 성능을 나타내고, 오전 9:30 및 오후 15:30 에 수행한 실험에서 상대적으로 우위에 있는 정확도를 얻을 수 있었다. 또한, MBTI의 ENFP에 속한 10명의 피험자의 정확도 평균이 73%로 유일하게 다른 15개 유형에 비하여 70%의 성능을 넘는 수준을 보여주는 것을 확인 할 수 있었다. Motor imagery is well used paradigm for Brain-Computer Interface (BCI) especially in 2 class BCI since it shows discriminative spatial pattern. But some do not perform well in such BCI system. These people are called BCI-illiteracy. Therefore it is important to estimate individual's potential performance in BCI system. In this study, we investigate the performance over various factor using 52 motor imagery datasets. As a result, many of datasets shows their performance near chance level (around 60% when 100trials for each condition). Female subjects perform better than male and some conducted in 9:30am and 15:30pm results relatively better accuracies. In addition, we found that the group in ENFP type in MBTI test shows its mean accuracy about 73% which is only one which reaches over 70% accuracy among 16 MBTI groups.