Image Captioning with Synergy-Gated Attention and Recurrent Fusion LSTM

Yang, You,Chen, Lizhi,Pan, Longyue,Hu, Juntao Korean Society for Internet Information 2022 KSII Transactions on Internet and Information Syst Vol.16 No.10

Long Short-Term Memory (LSTM) combined with attention mechanism is extensively used to generate semantic sentences of images in image captioning models. However, features of salient regions and spatial information are not utilized sufficiently in most related works. Meanwhile, the LSTM also suffers from the problem of underutilized information in a single time step. In the paper, two innovative approaches are proposed to solve these problems. First, the Synergy-Gated Attention (SGA) method is proposed, which can process the spatial features and the salient region features of given images simultaneously. SGA establishes a gated mechanism through the global features to guide the interaction of information between these two features. Then, the Recurrent Fusion LSTM (RF-LSTM) mechanism is proposed, which can predict the next hidden vectors in one time step and improve linguistic coherence by fusing future information. Experimental results on the benchmark dataset of MSCOCO show that compared with the state-of-the-art methods, the proposed method can improve the performance of image captioning model, and achieve competitive performance on multiple evaluation indicators.

A Robust Object Detection and Tracking Method using RGB-D Model

박서희,전준철,Park, Seohee,Chun, Junchul Korean Society for Internet Information 2017 인터넷정보학회논문지 Vol.18 No.4

Recently, CCTV has been combined with areas such as big data, artificial intelligence, and image analysis to detect various abnormal behaviors and to detect and analyze the overall situation of objects such as people. Image analysis research for this intelligent video surveillance function is progressing actively. However, CCTV images using 2D information generally have limitations such as object misrecognition due to lack of topological information. This problem can be solved by adding the depth information of the object created by using two cameras to the image. In this paper, we perform background modeling using Mixture of Gaussian technique and detect whether there are moving objects by segmenting the foreground from the modeled background. In order to perform the depth information-based segmentation using the RGB information-based segmentation results, stereo-based depth maps are generated using two cameras. Next, the RGB-based segmented region is set as a domain for extracting depth information, and depth-based segmentation is performed within the domain. In order to detect the center point of a robustly segmented object and to track the direction, the movement of the object is tracked by applying the CAMShift technique, which is the most basic object tracking method. From the experiments, we prove the efficiency of the proposed object detection and tracking method using the RGB-D model. 최근 지능형 CCTV는 빅 데이터, 인공지능 및 영상 분석과 같은 분야와 결합하여 다양한 이상 행위들을 탐지하고 보행자와 같은 객체의 전반적인 상황을 분석할 수 있으며, 이러한 지능형 영상 감시 기능에 대한 영상 분석 연구가 활발히 진행되고 있는 추세이다. 그러나 일반적으로 2차원 정보를 이용하는 CCTV 영상은 위상학적 정보 부족으로 인해 객체 오 인식과 같은 한계가 존재한다. 이러한 문제는 두 대의 카메라를 사용하여 생성된 객체의 깊이 정보를 영상에 추가함으로써 해결 할 수 있다. 본 논문에서는 가우시안 혼합기법을 사용하여 배경 모델링을 수행하고, 모델링 된 배경에서 전경을 분할하여 움직이는 객체의 존재 여부를 탐지한다. RGB 정보 기반 분할 결과를 이용하여 깊이 정보 기반 분할을 수행하기 위해 두 대의 카메라를 사용하여 스테레오 기반 깊이 지도를 생성한다. RGB 기반으로 분할된 영역을 깊이 정보를 추출하기 위한 도메인으로 설정하고, 도메인 내부에서 깊이 기반 분할을 수행한다. 강건하게 분할된 객체의 중심점을 탐지하고 방향을 추적하기 위해 가장 기본적인 객체 추적 방법인 CAMShift 기법을 적용하여 객체의 움직임을 추적한다. 실험을 통하여 제안된 RGB-D 모델을 이용한 객체 탐지 및 추적 방법의 우수성을 입증하였다.

Information-Sharing Patterns of A Directed Social Network: The Case of Imhonet

Lee, Danielle Korean Society for Internet Information 2017 인터넷정보학회논문지 Vol.18 No.4

Despite various types of online social networks having different topological and functional characteristics, the kinds of online social networks considered in social recommendations are highly restricted. The pervasiveness of social networks has brought scholarly attention to expanding the scope of social recommendations into more diverse and less explored types of online social networks. As a preliminary attempt, this study examined the information-sharing patterns of a new type of online social network - unilateral (directed) network - and assessed the feasibility of the network as a useful information source. Specifically, this study mainly focused on the presence of shared interests in unilateral networks, because the shared information is the inevitable condition for utilizing the networks as a feasible source of personalized recommendations. As the results, we discovered that user pairs with direct and distant links shared significantly more similar information than the other non-connected pairs. Individual users' social properties were also significantly correlated with the degree of their information similarity with social connections. We also found the substitutability of online social networks for the top cohorts anonymously chosen by the collaborative filtering algorithm.

Applying the IoT platform and green wave theory to control intelligent traffic lights system for urban areas in Vietnam

Phan, Cao Tho,Pham, Duy Duong,Tran, Hoang Vu,Tran, Trung Viet,Huu, Phat Nguyen Korean Society for Internet Information 2019 KSII Transactions on Internet and Information Syst Vol.13 No.1

This paper proposes an intelligent system performing an application with assistance of an Internet of Things (IoT) platform to control a traffic lights system. In our proposed systems, the traffic lights can be remotely controlled through the Internet. Based on IoT platform, the traffic conditions at different intersections of roads are collected and the traffic lights are controlled in a central manner. For the software part, the algorithm is designed based on the green wave theory to maximize the green bandwidth of arterial roads while addressing a challenging issue: the rapid changes of parameters including cycle time, splits, offset, non-fixed vehicles' velocities and traffic flow along arterial roads. The issue typically happens at some areas where the transportation system is not well organized like in Vietnam. For the hardware part, PLC S7-1200 are placed at the intersections for two purposes: to control traffic lights and to collect the parameters and transmit to a host machine at the operation center. For the communication part, the TCP/IP protocol can be done using a Profinet port embedded in the PLC. Some graphical user interface captures are also presented to illustrate the operation of our proposed system.

R3: A Lightweight Reactive Ring based Routing Protocol for Wireless Sensor Networks with Mobile Sinks

Yu, Sheng,Zhang, Baoxian,Yao, Zheng,Li, Cheng Korean Society for Internet Information 2016 KSII Transactions on Internet and Information Syst Vol.10 No.12

Designing efficient routing protocols for a wireless sensor network with mobile sinks (mWSN) is a challenging task since the network topology and data paths change frequently as sink nodes move. In this paper, we design a novel lightweight reactive ring based routing protocol called R3, which removes the need of proactively maintaining data paths to mobile sinks as they move in the network. To achieve high packet delivery ratio and low transmission cost, R3 combines ring based forwarding and trail based forwarding together. To support efficient ring based forwarding, we build a ring based structure for a network in a way such that each node in the network can easily obtain its ring ID and virtual angle information. For this purpose, we artificially create a virtual hole in the central area of the network and accordingly find a shortest cycled path enclosing the hole, which serves as base ring and is used for generating the remaining ring based structure. We accordingly present the detailed design description for R3, which only requires each node to keep very limited routing information. We derive the communication overhead by ring based forwarding. Extensive simulation results show that R3 can achieve high routing performance as compared with existing work.

Communication Strategies of Online-Based Leadership and Members' Work Engagement and Job Burnout

함상우,Hahm, Sang-Woo Korean Society for Internet Information 2017 인터넷정보학회논문지 Vol.18 No.5

Work engagement and job burnout are crucial problems in Korea, particularly in the business environment, and they also indicate people's motivation. Online-based teams in e-business also have these issues. Especially the role of the leader in this new type of teams is more crucial for members' motivation, behavior, and performance. Contingency leadership theories posit that different situations need different traits or behaviors of a leader. For an online-based team, which is internet-based team in e-business, a leader should use specific communication strategies to enhance a team's performance and members' motivation. In an online-based team and an offline-based team, the ways in which leaders motivate members will be different. This paper suggests particular communicative strategies for leaders of online situations. Online-based team members communicate using mobile phones or IT tools, and leaders mayinfluences members in virtual meetings. Moreover, online-based leadership needs to be more entrepreneurial and risk encouraging, and less conservative than offline-based leadership. In addition, online-based leadership should be energetic, decisive, adaptable, and inspiring, and furtherutilize communication tools such as being more gregarious/talkative, assertiveness, linguistic diversity and affect. This paper explains the roles of communicative elements such as coordination, expression, direction-giving/uncertainty reduction, empathy, meaning-making, satisfaction, and competence to decrease job burnout and to increase work engagement. The purpose of this study is to suggest communication strategies' differentiated influences on work engagement and job burnout in an online-based team.

An Efficient Multicast-based Binding Update Scheme for Network Mobility

Kim, Moon-Seong,Radha, Hayder,Lee, Jin-Young,Choo, Hyun-Seung Korean Society for Internet Information 2008 KSII Transactions on Internet and Information Syst Vol.2 No.1

Mobile IP (MIP) is the solution supporting the mobility of Mobile Nodes (MNs), however, it is known to lack the support for NEtwork MObility (NEMO). NEMO manages situations when an entire network, composed of one or more subnets, dynamically changes its point of attachment to the Internet. NEMO Basic Support (NBS) protocol ensures session continuity for all the nodes in a mobile network, however, there exists a serious pinball routing problem. To overcome this weakness, there are many Route Optimization (RO) solutions such as Bi-directional Tunneling (BT) mechanism, Aggregation and Surrogate (A&S) mechanism, Recursive Approach, etc. The A&S RO mechanism is known to outperform the other RO mechanisms, except for the Binding Update (BU) cost. Although Improved Prefix Delegation (IPD) reduces the cost problem of Prefix Delegation (PD), a well-known A&S protocol, the BU cost problem still presents, especially when a large number of Mobile Routers (MRs) and MNs exist in the environment such as train, bus, ship, or aircraft. In this paper, a solution to reduce the cost of delivering the BU messages is proposed using a multicast mechanism instead of unicasting such as the traditional BU of the RO. The performance of the proposed multicast-based BU scheme is examined with an analytical model which shows that the BU cost enhancement is up to 32.9% over IPDbased, hence, it is feasible to predict that the proposed scheme could benefit in other NEMO RO protocols.

Relaying Protocols and Delay Analysis for Buffer-aided Wireless Powered Cooperative Communication Networks

Zhan, Jun,Tang, Xiaohu,Chen, Qingchun Korean Society for Internet Information 2018 KSII Transactions on Internet and Information Syst Vol.12 No.8

In this paper, we investigate a buffer-aided wireless powered cooperative communication network (WPCCN), in which the source and relay harvest the energy from a dedicated power beacon via wireless energy transfer, then the source transmits the data to the destination through the relay. Both the source and relay are equipped with an energy buffer to store the harvested energy in the energy transfer stage. In addition, the relay is equipped with a data buffer and can temporarily store the received information. Considering the buffer-aided WPCCN, we propose two buffer-aided relaying protocols, which named as the buffer-aided harvest-then-transmit (HtT) protocol and the buffer-aided joint mode selection and power allocation (JMSPA) protocol, respectively. For the buffer-aided HtT protocol, the time-averaged achievable rate is obtained in closed form. For the buffer-aided JMSPA protocol, the optimal adaptive mode selection scheme and power allocation scheme, which jointly maximize the time-averaged throughput of system, are obtained by employing the Lyapunov optimization theory. Furthermore, we drive the theoretical bounds on the time-averaged achievable rate and time-averaged delay, then present the throughput-delay tradeoff achieved by the joint JMSPA protocol. Simulation results validate the throughput performance gain of the proposed buffer-aided relaying protocols and verify the theoretical analysis.

An Analysis of Card News and Deconstructing News Values in Curated News Contents in the Digital Era

홍성철,배정근,Hong, Seong Choul,Pae, Jung Kun Korean Society for Internet Information 2017 인터넷정보학회논문지 Vol.18 No.4

본 연구는 새로운 뉴스 콘텐츠로서 카드뉴스를 분석하고 디지털시대에 뉴스가치의 변화에 대해 살펴보았다. 본 연구는 지난 2015년 하반기에 국내 언론사를 통해 보도된 1020개의 카드뉴스에 대한 내용분석을 통해, 뉴스가치에 대한 전통적인 개념들이 변화되고 있음을 확인하였다. 카드뉴스는 새롭게 취재하여 기사화하기보다는 저장된 기존 뉴스들을 재활용하면서, 뉴스가치로서 시의성 보다는 정보제공과 사회적 가치, 오락적 요소를 강조하고 있다. 이는 새로운 뉴스전달방식으로써 카드뉴스는 기존 뉴스에 대한 대체재가 아니라 보완재임을 보여준다. 또한 역피라밋 형태의 글쓰기보다는 리스트 나열형의 기사 작성 방식이 선호되고, 취재원으로서 길거리 일반인(Man-on-the streets)이 자주 인용됨이 발견되었다. This paper explores the characteristics of curated news content. With content analysis of 1020 news clips, the study found that news values immersed in card news differed from those of traditional news. Specifically, timeliness was not regarded as a key factor in newsworthiness. Rather, information and social impacts were highly emphasized. Considering news consumers depend on traditional news for timely news, curated news content was not a replacement for traditional news but a supplement. By refurbishing photos from previous news reports and googling the web for related information, curated news reiterates social meaning and provides relevant information. Furthermore, salience of human interest can be explained by entertaining characteristics of curated news. In story forms, the list technique has several important points to stress, and was more frequently used than inverted pyramids. Another key finding of this study is man-on-the street as the most quoted main sources in the curatorial context.

Construction of an PFT database with various clinical information using optical character recognition and regular expression technique

Park, Man Young,Park, Rae Woong Korean Society for Internet Information 2017 인터넷정보학회논문지 Vol.18 No.5

The pulmonary function test (PFT) is an essential data source for evaluating the effect of drugs on the lungs or the status of lung function. However, the numeric values of PFT cannot be easily used for clinical studies without labor-intensive manual efforts, because PFTs are usually recorded as image files. This study was aimed at constructing a de-identified, open-access PFT database with various clinical information. For constructing the PFT database, optical character recognition (OCR), regular expression, and the parsing technique were used to extract alphanumeric data from the PFT images in a Korean tertiary teaching hospital. This longitudinal observational database contains 413,000 measurements of PFT from 183,000 patients.