Search

Deep learning technology has rapidly evolved in recent years. Bone age assessment (BAA) is a typical object detection and classification problem that would benefit from deep learning. Convolutional neural networks (CNNs) and their variants are hence increasingly used for automating BAA, and they have shown promising results. In this paper, we propose a complete end-to-end BAA system to automate the entire process of the Tanner-Whitehouse 3 method, starting from localization of the epiphysis-metaphysis growth regions within 13 different bones and ending with estimation of the corresponding BA. Specific modifications to the CNNs and other stages are proposed to improve results. In addition, an annotated database of 3300 X-ray images is built to train and evaluate the system. The experimental results show that the average top-1 and top-2 prediction accuracies for skeletal bone maturity levels for 13 regions of interest are 79.6% and 97.2%, respectively. The mean absolute error and root mean squared error in age prediction are 0.46 years and 0.62 years, respectively, and accuracy within one year of the ground truth of 97.6% is achieved. The proposed system is shown to outperform a commercially available Greulich-Pyle-based system, demonstrating the potential for practical clinical use.

Moving pictures experts group (MPEG) media transport (MMT) and MPEG-dynamic adaptive streaming over hypertext transfer protocol (MPEG-DASH) are emerging as the content-delivery technologies for the next-generation broadcasting systems, as they seamlessly utilize various underlying delivery networks to provide hybrid multimedia services in any location. In this paper, we present an empirical analysis of the channel-zapping time, which is an important metric in the measurement of the quality of experience (QoE) of broadcast services, regarding the MMT and MPEG-DASH technologies. During this process, we also clarify the important factors that determine the channel-zapping time of the individual technologies. In addition, we propose a simple yet effective method that enables fast channel zapping for the MMT technology.

The objective of this paper is to improve the overall performance of distributed encoding, with a particular focus on encoding speed. The proposed scheme consists of content-aware video segmentation and scheduling that consists of two major parts. In the first part, segmentation is carried out with greater efficiency by considering changes in the video content, and in the second part, the segment assignment process is carried out using an efficient scheduling scheme that changes the encoding order of the segments. We measured the content similarity by using the sum of absolute difference algorithm and then applied a threshold to define the degree of change in similarity. The video was segmented based on the extent to which the similarity had changed, and the encoding order of the segments was rearranged to perform distributed encoding. Finally, this paper introduces the MapReduce-based distributed video encoding, using the content-aware video segmentation and scheduling described above, and presents the results of the performance using this scheme, which indicate that the proposed scheme increases the bitrate by a maximum of 2.9% over existing segmentation schemes, and also increases the speed by a maximum of 15.3%.

With the recent advancements in cloud computing technology, the number of cloud-based services has been gradually increasing. Symmetrically, users are asking for quality of experience (QoE) to be maintained or improved. To do this, it has become necessary to manage network resources more efficiently inside the cloud. Many theoretical studies for improving the users’ QoE have been proposed. However, there are few practical solutions due to the lack of symmetry between implementation and theoretical researches. Hence, in this study, we propose a ranking table-based network resource allocation method that dynamically allocates network resources per service flow based on flow information periodically collected from a software defined network (SDN). It dynamically identifies the size of the data transmission for each service flow on the SDN and differentially allocates network resources to each service flow based on this size. As a result, it maintains the maximum QoE for the user by increasing the network utilization. The experimental results show that the proposed method achieves 29.4% higher network efficiency than the general Open Shortest Path First (OSPF) method on average.

The Software Defined Network (SDN) decouples control and data planes of the existing network architecture and a controller carries out centralized management and control of the network. This provides flexibility for the networks and enables efficient operation, depending on their condition. However, when the performance of switches comprising the SDN varies, any switches with low performance can cause flow table overflow and subsequent bottleneck with overhead concentrating in the controller. Besides, packet processing can be delayed, declining user QoS. The paper proposes a flow entry replacement scheme that takes into account the flow features to address the issue. The proposed Short Flow First (SFF) scheme discriminates the flow features based on the matching period of flow entry and determines a subject of the flow entry replacement with the features. For evaluation, the paper built the SDN environment using the Mininet emulator and compared the performance of the proposed SFF scheme with the representative flow entry replacement schemes. According to the results, the proposed scheme showed a good performance in the flow entry miss rate and packet delay due to failure of matching the flow entry.

In wireless ad hoc networks, energy-constrained operation is very important due to the limited battery capacity. One of the reasons for excessive energy expenditure in this network is irregularly distributed node patterns which impose large interference range on a certain area. In this paper, we present a simple and energy efficient on-demand routing scheme by using discrete level of power control and priority based packet scheduling. With little additional routing overhead, the proposed scheme provides better way to deliver data than existing routing protocols which use fixed transmission power.

In order to build an efficient security architecture, previous studies have attempted to understand complex system architectures and message flows to detect various attack packets. However, the existing hardware-based single security architecture cannot efficiently handle a complex system structure. To solve this problem, we propose a software-defined networking (SDN) policy-based scheme for an efficient security architecture. The proposed scheme considers four policy functions: separating, chaining, merging, and reordering. If SDN network functions virtualization (NFV) system managers use these policy functions to deploy a security architecture, they only submit some of the requirement documents to the SDN policy-based architecture. After that, the entire security network can be easily built. This paper presents information about the design of a new policy functions model, and it discusses the performance of this model using theoretical analysis.

In wireless body sensor network systems (WB-SNSs), the sensor nodes have very limited battery power because they are tiny, lightweight, and wearable or implantable. As a result, WB-SNSs require a very efficient transmission power control (TPC) algorithm for effectively reducing energy consumption and extending the lifetime of sensor nodes. To achieve this goal, we propose a novel TPC algorithm referred to as hybrid TPC. The hybrid TPC algorithm adaptively selects a conservative or an aggressive control mechanism depending on current channel conditions. The conservative control mechanism, which slowly changes transmission power level (TPL), is suitable in a dynamic environment. On the other hand, the aggressive control mechanism, which rapidly changes TPL, is ideal in a static environment. In order to evaluate the effectiveness of the hybrid TPC algorithm, we implemented various TPC algorithms and compared their performances against the hybrid TPC algorithm in different channel environments. The experimental results showed that the hybrid TPC algorithm outperformed other TPC algorithms in all channel environments.

In order to construct an efficient wireless sensor system, it is necessary to increase the lifetime of its battery-operated sensor nodes. To this end, a wireless body sensor system adopts a transmission power control (TPC) mechanism. However, existing TPC mechanisms adjust the transmission power level (TPL) according to the received signal strength indication (RSSI) value of the most recently received data packet. Therefore, they do not effectively cope with dynamically changing wireless body channel environments. In particular, when a wireless channel is unstable, changes in the TPL should be avoided to prevent energy consumption due to unnecessary transmission and reception of control packets. Accordingly, this paper proposes a new TPC mechanism that adaptively changes the control packet transmission interval on the basis of the current channel condition. Further, actual sensors are used to experimentally verify that the proposed mechanism (1) performs well in all channel environments and (2) facilitates the construction of an efficient wireless body sensor system.

This research aims to propose a practical framework designed for the automatic analysis of a product's comprehensive functionality and security vulnerabilities, generating applicable guidelines based on real-world software. The existing analysis of software security vulnerabilities often focuses on specific features or modules. This partial and arbitrary analysis of the security vulnerabilities makes it challenging to comprehend the overall security vulnerabilities of the software. The key novelty lies in overcoming the constraints of partial approaches. The proposed framework utilizes data from various sources to create a comprehensive functionality profile, facilitating the derivation of real-world security guidelines. Security guidelines are dynamically generated by associating functional security vulnerabilities with the latest Common Vulnerabilities and Exposure (CVE) and Common Vulnerability Scoring System (CVSS) scores, resulting in automated guidelines tailored to each product. These guidelines are not only practical but also applicable in real-world software, allowing for prioritized security responses. The proposed framework is applied to virtual private network (VPN) software, wherein a validated Level 2 data flow diagram is generated using the Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of privilege (STRIDE) technique with references to various papers and examples from related software. The analysis resulted in the identification of a total of 121 vulnerabilities. The successful implementation and validation demonstrate the framework's efficacy in generating customized guidelines for entire systems, subsystems, and selected modules. [ABSTRACT FROM AUTHOR]

Recommendation Information Systems (RIS) are pivotal in helping users in swiftly locating desired content from the vast amount of information available on the Internet. Graph Convolution Network (GCN) algorithms have been employed to implement the RIS efficiently. However, the GCN algorithm faces limitations in terms of performance enhancement owing to the due to the embedding value-vanishing problem that occurs during the learning process. To address this issue, we propose a Weighted Forwarding method using the GCN (WF-GCN) algorithm. The proposed method involves multiplying the embedding results with different weights for each hop layer during graph learning. By applying the WF-GCN algorithm, which adjusts weights for each hop layer before forwarding to the next, nodes with many neighbors achieve higher embedding values. This approach facilitates the learning of more hop layers within the GCN framework. The efficacy of the WF-GCN was demonstrated through its application to various datasets. In theMovieLens dataset, the implementation of WF-GCN in LightGCN resulted in significant performance improvements, with recall and NDCG increasing by up to +163.64% and +132.04%, respectively. Similarly, in the Last. FM dataset, LightGCN using WF-GCN enhanced with WF-GCN showed substantial improvements, with the recall and NDCG metrics rising by up to +174.40% and +169.95%, respectively. Furthermore, the application of WF-GCNto Self-supervisedGraph Learning (SGL) and SimpleGraph Contrastive Learning (SimGCL) also demonstrated notable enhancements in both recall and NDCG across these datasets. [ABSTRACT FROM AUTHOR]