Robust person re-identification via graph convolution networks

Person re-identification (re-id) aims to identity the same person over multiple cameras; it has been successfully applied to various computer vision applications as a fundamental method. Owing to the development of deep learning, person re-id methods, which typically use triplet networks based on triplet loss, have demonstrated great success. However, the appearances of people are similar and hence difficult to distinguish in many cases. Therefore, we present a novel graph convolution network and enhances traditional triplet loss functions. Our method defines reference, positive, and negative features for triplet loss as three vertices of a graph, respectively, and adjusts their mutual distance through learning. The method adopts graph convolutions efficiently, thereby affording low computational costs. Experimental results demonstrate that our method is superior to the baseline on the Market-1501 dataset. The proposed GCN-based triplet loss considerably contributes to improve re-identification methods quantitatively and qualitatively.

     

A resource allocation policy for delay minimization in fetching capacitated feeds

As social media services such as Twitter and Facebook are gaining popularity, the amount of information published from those services is explosively growing. Most of them use feeds to facilitate distribution of a huge volume of content they publish. In this context, many users subscribe to feeds to acquire up-to-date information through feed aggregation services, and recent real-time search engines also increasingly utilize feeds to promptly find recent web content when it is produced. Accordingly, it is necessary for such services to effectively fetch feeds for minimizing fetching delay, while at the same time maximizing the number of fetched entries. Fetching delay is a time lag between entry publication and retrieval, which is primarily incurred by finiteness of fetching resources. In this paper, we consider a polling-based approach among the methods applicable to fetching feeds, which bases on a specific schedule for visiting feeds. While the existing polling-based approaches have focused on the allocation of fetching resources to feeds in order to either reduce the fetching delay or increase the number of fetched entries, we propose a resource allocation policy that can optimize both objectives. Extensive experiments have been carried out to evaluate the proposed model, in comparison with the existing alternative methods.     

Microfluidic-based biosensors toward point-of-care detection of nucleic acids and proteins

This article reviews state-of-the-art microfluidic biosensors of nucleic acids and proteins for point-of-care (POC) diagnostics. Microfluidics is capable of analyzing small sample volumes (10⁻⁹–10⁻¹⁸ l) and minimizing costly reagent consumption as well as automating sample preparation and reducing processing time. The merger of microfluidics and advanced biosensor technologies offers new promises for POC diagnostics, including high-throughput analysis, portability and disposability. However, this merger also imposes technological challenges on biosensors, such as high sensitivity and selectivity requirements with sample volumes orders of magnitude smaller than those of conventional practices, false response errors due to non-specific adsorption, and integrability with other necessary modules. There have been many prior review articles on microfluidic-based biosensors, and this review focuses on the recent progress in last 5 years. Herein, we review general technologies of DNA and protein biosensors. Then, recent advances on the coupling of the biosensors to microfluidics are highlighted. Finally, we discuss the key challenges and potential solutions for transforming microfluidic biosensors into POC diagnostic applications.     

A monolithic three-axis micro-g micromachined silicon capacitive accelerometer

A monolithic three-axis micro-g resolution silicon capacitive accelerometer system utilizing a combined surface and bulk micromachining technology is demonstrated. The accelerometer system consists of three individual single-axis accelerometers fabricated in a single substrate using a common fabrication process. All three devices have 475-/spl mu/m-thick silicon proof-mass, large area polysilicon sense/drive electrodes, and small sensing gap (<1.5 /spl mu/m) formed by a2004 sacrificial oxide layer. The fabricated accelerometer is 7/spl times/9 mm/sup 2/ in size, has 100 Hz bandwidth, >/spl sim/5 pF/g measured sensitivity and calculated sub-/spl mu/g//spl radic/Hz mechanical noise floor for all three axes. The total measured noise floor of the hybrid accelerometer assembled with a CMOS interface circuit is 1.60 /spl mu/g//spl radic/Hz (>1.5 kHz) and 1.08 /spl mu/g//spl radic/Hz (>600 Hz) for in-plane and out-of-plane devices, respectively.     

A scalable model for interbandwidth broker resource reservation and provisioning

As the Internet evolves into global communication and commercial infrastructure, the need for quality-of-services (QoSs) in the Internet becomes more and more important. With a bandwidth broker (BB) support in each administrative domain, differentiated services (Diffserv) is seen as a key technology for achieving QoS guarantees in a scalable, efficient, and deployable manner in the Internet. In this paper, we present a scalable model for inter-BB resource reservation and provisioning. Our BB uses centralized network state maintenance and pipe-based intradomain resource management schemes that significantly reduce admission control time and minimize scalability problems present in prior research. For inter-BB communication, we design and implement a BB resource reservation and provisioning protocol (BBRP). BBRP performs destination-based aggregated resource reservation based on bilateral service level agreements (SLAs) between peer-BBs. BBRP significantly reduces the BB and border routers state scalability problem by maintaining reservation state based only on destination region. It minimizes inter-BB signaling scalability by using aggregated type resource reservation and provisioning. Both analytical and experimental results verify the BBRP achievements.     

Design and evaluation of inter‐bandwidth broker signaling

Bandwidth brokers (BBs) have been proposed for providing end‐to‐end quality of services (QoS) in differentiated services (Diffserv) networks. As a single entity in each domain, a BB aims at performing both intra‐ and inter‐domain resource management on behalf of its domain. There have been plenty of BB studies for intra‐domain resource management. However, how a BB can perform inter‐domain resource management in a scalable and deployable manner is still an open issue. In this work, we present the design, implementation and evaluation of an inter‐BB communication protocol that is used by each BB to communicate with its neighboring BBs for inter‐domain QoS resource management. The proposed model uses a destination‐based aggregation scheme in which reservations are aggregated as they merge through the destination region. The destination‐based aggregation improves inter‐domain state and signaling scalability. The protocol also performs traffic engineering to increase inter‐domain resource utilization. The implementation and simulation results verify the achievements of our model. Copyright © 2008 John Wiley & Sons, Ltd.     

Modeling occupant behavior of the manual control of windows in residential buildings

Window opening and closing is the most preferred behavior for occupants to control their indoor environment in homes. This study aims to identify driving forces for window opening and closing behavior in the home. The additional field survey was carried out for the cooling period after following the previous study. The state of windows and environmental variables for outdoor and indoor were continuously monitored in 23 sample homes over one year. The monitored data provide evidence that there is a statistically significant relationship between window opening behavior and outdoor temperature. The behavior of the occupant's manual control of windows can be described by seasonal effects, occupancy, and time of day. Indoor stimuli, such as such as temperature, humidity, and CO₂, can better account for the window opening behavior than can outdoor stimuli. There are clear differences in driving variables between window opening and closing behavior. The closing behavior is better described when the outdoor and indoor variables are combined. Finally, multivariate logistic regression models were developed to predict typical patterns of window opening and closing as a function of indoor and outdoor variables.     

A Projection Method for the Conservative Discretizations of Parabolic Partial Differential Equations

We present a projection method for the conservative discretizations of parabolic partial differential equations. When we solve a system of discrete equations arising from the finite difference discretization of the PDE, we can use iterative algorithms such as conjugate gradient, generalized minimum residual, and multigrid methods. An iterative method is a numerical approach that generates a sequence of improved approximate solutions for a system of equations. We repeat the iterative algorithm until a numerical solution is within a specified tolerance. Therefore, even though the discretization is conservative, the actual numerical solution obtained from an iterative method is not conservative. We propose a simple projection method which projects the non-conservative numerical solution into a conservative one by using the original scheme. Numerical experiments demonstrate the proposed scheme does not degrade the accuracy of the original numerical scheme and it preserves the conservative quantity within rounding errors.