• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.7
• /
• pp.2407-2424
• /
• 2022

Recently, Transformer has made great progress in video retrieval tasks due to its high representation capability. For the structure of a Transformer, the cascaded self-attention modules are capable of capturing long-distance feature dependencies. However, the local feature details are likely to have deteriorated. In addition, increasing the depth of the structure is likely to produce learning bias in the learned features. In this paper, an improved Transformer structure named TransDCS (Transformer with Dynamic Convolution and Shortcut) is proposed. A Multi-head Conv-Self-Attention module is introduced to model the local dependencies and improve the efficiency of local features extraction. Meanwhile, the augmented shortcuts module based on a dual identity matrix is applied to enhance the conduction of input features, and mitigate the learning bias. The proposed model is tested on MSRVTT, LSMDC and Activity-Net benchmarks, and it surpasses all previous solutions for the video-text retrieval task. For example, on the LSMDC benchmark, a gain of about 2.3% MdR and 6.1% MnR is obtained over recently proposed multimodal-based methods.

• The Journal of Korean Physical Therapy
• /
• v.17 no.3
• /
• pp.351-367
• /
• 2005

The purpose of this study was to identity effects of virtual reality(VR) program related to standing postural control on balance, gait and brain activation patterns in chronic hemiplegic stroke patients. Subjects were assigned randomly to either VR group (n=12) or the control group (n=12) when the study began. Both groups received conventional physical therapy for 2 to 3 times per week. In addition to conventional physical therapies, VR group trained 3 types of virtual reality programs using IREX for standing postural control during 4 weeks (4 times/week, 30 minutes/time). Subjects were assessed for static and dynamic balance parameters using BPM, functional balance using Berg Balance Scale related to movement of paretic lower limb before and after 4 weeks of virtual reality training. The results of this study were as follows. 1. Following VR training, VR group demonstrated the marked improvement on dynamic mean balance, anteroposterior limits of stability (AP angle) and mediolateral limits of stability (ML angle). 2. Following VR training, both groups scored higher on Berg Balance Scale. However, a comparison of mean change revealed differences between groups. In conclusion, these data suggest that the postural control training using VR programs improve dynamic and functional balance performance in chronic hemiplegic stroke patients.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.13 no.1
• /
• pp.87-101
• /
• 2017

As people increasingly rely on mobile networks in modern society, mobile communication security is becoming more and more important. In the Long Term Evolution/System Architecture Evolution (LTE/SAE) architecture, the 3rd Generation Partnership (3GPP) team has also developed the improved Evolved Packet System Authentication and Key Agreement (EPS AKA) protocol based on the 3rd Generation Authentication and Key Agreement (3G AKA) protocol in order to provide mutual authentication and secure communication between the user and the network. Unfortunately, the EPS AKA also has several vulnerabilities such as sending the International Mobile Subscriber Identity (IMSI) in plain text (which leads to disclosure of user identity and further causes location and tracing of the user, Mobility Management Entity (MME) attack), man-in-middle attack, etc. Hence, in this paper, we analyze the EPS AKA protocol and point out its deficiencies and then propose an Efficient and Security Enhanced Authentication and Key agreement (ESE-EPS AKA) protocol based on hybrid of Dynamic Pseudonym Mechanism (DPM) and Public Key Infrastructure (PKI) retaining the original framework and the infrastructure of the LTE network. Then, our evaluation proves that the proposed new ESE-EPS AKA protocol is relatively more efficient, secure and satisfies some of the security requirements such as confidentiality, integrity and authentication.

• The Journal of the Acoustical Society of Korea
• /
• v.11 no.2
• /
• pp.68-72
• /
• 1992

Recently, many researchers have studied the speaker recognition through the statistical processing method using Karhunen-Loeve Transform. However, the content of speaker's identity and the vocalization speed cause speaker recognition rate to be lowered. This parer studies the speaker recognition method using weighted orthogonal parameters which are weighted with eigen-values of speech so as to emphasize the speaker's identity, and optimal path which is made by DWP so as to normalize dynamic time feature of speech. To confirm this method, we compare the speaker recognition rate from this proposed method with that from the conventional statistical processing method. As a result, it is shown that this method is more excellent in speaker recognition rate than conventional method.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.29 no.5
• /
• pp.60-70
• /
• 2001

This landscape design proposal was presented to a design competition for Gwangjin-gu Community Center and Assembly Hall. The site is located in 227-7 Jayang-dong, Gwangjin-gu, Seoul and is part a block designated as \`Kunkuk University Detailed Planning District\`. The judging criteria for landscape design set by the Gwangjin-gu could be articulated as follows: an identity of Gwangjin-gu, a relationship with surrounding environment, a locale as a community center of Gwangjin-gu, and an environment-friendly design. An identity of Gwangjin-gu was expressed by introducing lends and decks around the building which were transformed from the image of the old Gwangjin Ferry. The concept\` was strengthened by \`Nori-madang\` or Korean version of \`plaza\` which was designed by a motive of \`mungsuk\` or Korean old fashioned straw mat. An ambiguous relationship between the site and the surroundings was set by \`transparent exhibition frame\` which not only bleaks the massive Chungdam-daegyo, an elevated road, and frames it into small parts, but also attracts pedestrians toward the site-the \`park\`. Pedestrians from the adjacent\` Nengdong-ro Walkable Streets\` views the site through the many parts of frames. By introducing diverse resting and dynamic spaces for many kinds of cultural activities the site could function as a locale of the Gwangjin-gu community. The main paving system reflects the environment- friendly design. It was detailed by \`PC-block-deck-paving\` which was elevated from the ground so that rains could be absorbed.

• Journal of Communications and Networks
• /
• v.10 no.1
• /
• pp.79-88
• /
• 2008

In this paper, we propose efficient peer assignment algorithms for low-latency transmission of scalable coded images in peer-to-peer networks, in which peers may dynamically join and leave the networks. The objective of our algorithm is to minimize the transmission time of a requested image that is scalable coded. When an image is scalable coded in different bit rates, the bit stream encoded in a lower bit rate is a prefix subset of the one encoded in a higher bit rate. Therefore, a peer with the same requested image coded in any bit rate, even when it is different from the requested rate, may work as a supplying peer. As a result, when a scalable coded image is requested, more supplying peers can be found in peer-to-peer networks to help with the transfer. However, the set of supplying peers is not static during transmission, as the peers in this set may leave the network or finish their transmission at different times. The proposed peer assignment algorithms have taken into account the above constraints. In this paper, we first prove the existence of an optimal peer assignment solution for a simple identity permutation function, and then formulate peer assignment with this identity permutation as a mixed-integer programming problem. Next, we discuss how to address the problem of dynamic peer departures during image transmission. Finally, we carry out experiments to evaluate the performance of proposed peer assignment algorithms.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.7
• /
• pp.1539-1544
• /
• 2003

Recently, many researchers have studied the speaker recognition through the statistical processing method using Karhonen-Loeve Transform. However, the content of speaker's identity and the vocalization speed cause speaker recognition rate to be lowered. This parer studies the speaker recognition method using weighted parameters which are weighted with eigen-values of speech so as to emphasize the speaker's identity and optimal path which is made by DWP so as to normalize dynamic time feature of speech. To confirm this method, we compare the speaker recognition rate from this proposed method with that from the conventional statistical processing method. As a result, it is shown that this method is more excellent in speaker recognition rate than conventional method.

• Korean Medical Education Review
• /
• v.23 no.2
• /
• pp.80-89
• /
• 2021

Professional identity formation (PIF) is an essential concept in professional education. Many scholars have explored conceptual frameworks of PIF and conducted empirical studies to advance an understanding of the construct in medical education. Despite its importance, it is unclear what educational approaches and assessment practices are actually implemented in medical education settings. Therefore, we conducted a literature review of empirical studies reporting educational practices for medical learners' PIF. We searched the Web of Science database using keywords and chose 37 papers for analysis based on inclusion and exclusion criteria. Thematic analysis was conducted. Most empirical papers (92%) were from North America and Western Europe and used qualitative research methods, including mixed methods (99%). The papers reported the use of reflection activities and elective courses for specific purposes, such as art as an educational activity. Patient and healthcare experiences were also found to be a central theme in medical learners' PIF. Through an iterative analysis of the key themes that emerged from the PIF studies, we derived the following key concepts and implications: (1) the importance of creating informal and incidental learning environments, (2) ordinary yet authentic patient experiences, (3) a climate of psychosocial safety in a learning environment embracing individual learners' background and emotional development, and (4) the reconceptualization of PIF education and assessment. In conclusion, research on PIF should be diversified to include various cultural and social contexts. Theoretical frameworks should also be diversified and developed beyond Kegan's developmental framework to accommodate the nonlinear and dynamic nature of PIF.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.39 no.1
• /
• pp.227-237
• /
• 2019

In terms of the concept of space in urban planning and design, the modern city approaches from the perspective that all spaces are made up of different places based on identity. In this study, we analyzed disparate image by paying attention to the difference between the individual spaces that form the city. And the physical and behavioral factors that respond to the disparate spatial images are derived through legibility measurement. As a result, the urban space in Busan has disparate images such as closure, openness, dynamic + closure, dynamic + openness, quiet space. The following are the legibility factors that make each image stronger. The elements of closure are high-density and high-rise buildings, residence, work, openness appeared as a factor of water space, walking and viewing. Dynamic & Closure is the element of industrial activity, dynamic & openness is open space, water space, swimming element, quiet space image is green space, walking, stay factor.

• Structural Engineering and Mechanics
• /
• v.32 no.3
• /
• pp.407-427
• /
• 2009

A fundamental solution for the transient, quasi-static, plane problems of linear viscoelasticity is introduced for a specific material. An integral equation has been found for any problem as a result of dynamic reciprocal identity which is written between this fundamental solution and the problem to be solved. The formulation is valid for the first, second and mixed boundary-value problems. This integral equation has been solved by BEM and algorithm of the BEM solution is explained on a sample, mixed boundary-value problem. The forms of time-displacement curves coincide with literature while time-surface traction curves being quite different in the results. The formulation does not have any singularity. Generalized functions and the integrals of them are used in a different form.