Image and Feature Space Based Domain Adaptation for Vehicle Detection

The application of deep learning in the field of object detection has experienced much progress. However, due to the domain shift problem, applying an off-the-shelf detector to another domain leads to a significant performance drop. A large number of ground truth labels are required when using another domain to train models, demanding a large amount of human and financial resources. In order to avoid excessive resource requirements and performance drop caused by domain shift, this paper proposes a new domain adaptive approach to cross-domain vehicle detection. Our approach improves the cross-domain vehicle detection model from image space and feature space. We employ objectives of the generative adversarial network and cycle consistency loss for image style transfer in image space. For feature space, we align feature distributions between the source domain and the target domain to improve the detection accuracy. Experiments are carried out using the method with two different datasets, proving that this technique effectively improves the accuracy of vehicle detection in the target domain.     

面向数据特征的人机物融合服务分派方法

随着工业互联网的不断发展，大数据和人工智能促成了人机物全面互联.用户使用服务时产生的任务数据量正呈指数级增长，在为线上用户推荐服务满足个性化需求的同时，对于需要通过人机物交互完成的服务，如何整合线上和线下资源，并分派合适的人快速、有效地完成任务，也已成为一个挑战性问题.为了保证服务分派的准确性，提出了一种综合考虑人机物各方面数据特征的跨域融合服务分派方法，分别对用户评价的情感倾向性和业务数据的相似性进行分析，然后加入对业务执行有影响的物理世界的属性特征，以获得更合理的分派.最后，以一个互联网在线诊疗平台的医患分派为例，结果表明，文中提出的分派方法具有较高的准确性，可以获得更好的用户体验.     

基于域对抗网络和BERT的跨领域文本情感分析

跨领域文本情感分析时,为了使抽取的共享情感特征能够捕获更多的句子语义信息特征,提出域对抗和BERT(bidirectional encoder representations from transformers)的深度网络模型。利用BERT结构抽取句子语义表示向量,通过卷积神经网络抽取句子的局部特征。通过使用域对抗神经网络使得不同领域抽取的特征表示尽量不可判别,即源领域和目标领域抽取的特征具有更多的相似性;通过在有情感标签的源领域数据集上训练情感分类器,期望该分类器在源领域和目标领域均能达到较好的情感分类效果。在亚马逊产品评论数据集上的试验结果表明,该方法具有良好的性能,能够更好地实现跨领域文本情感分类。     

5G端到端网络切片进展与挑战分析

随着5G端到端网络切片商用部署的逐步推进,在切片实施过程中遇到一些问题和挑战,因此亟需对其当前的研究进展及挑战进一步分析.从对5G端到端网络切片的系统性论述切入,首先介绍了5G网络切片的架构和功能,之后梳理了目前国内外标准组织对网络切片及其增强技术的研究进展,最后识别并深入分析了端到端网络切片面临的挑战,包括SLA保障...     

MPLS VPN跨域技术的分析与研究

介绍了几种跨域技术，详细地分析了各种技术的实现过程及其路由分发方式，并对各技术进行了总结和比较，给出了各自的适用条件。     

When Do Visual and Verbal Memories Conflict? The Importance of Working-Memory Load and Retrieval.

Examinations of interference between verbal and visual materials in working memory have produced mixed results. If there is a central form of storage (e.g., the focus of attention; N. Cowan, 2001), then cross-domain interference should be obtained. The authors examined this question with a visual-array comparison task (S. J. Luck & E. K. Vogel, 1997) combined with various verbal memory load conditions. Interference between tasks occurred if there was explicit retrieval of the verbal load during maintenance of a visual array. The effect was localized in the maintenance period of the visual task and was not the result of articulation per se. Interference also occurred when especially large silent verbal and visual loads were held concurrently. These results suggest central storage along with code-specific passive storage. (PsycINFO Database Record (c) 2011 APA, all rights reserved)     

Cross-domain access control policy mapping mechanism for balancing interoperability and autonomy

Cross-domain access control can improves interoperability but reduces intra-domain autonomy.To balance inter-domain interoperability and intra-domain autonomy,a cross-domain access control policy mapping to the problem of multi-objective integer optimization programming was formulated.Both the maximization of inter-domain interoperability and the minimization of intra-domain autonomy were taken as the objectives.Further,seven constraints were designed to prevent typical cross-domain conflicts.To solve the optimization problem,a constrained NSGA-III algorithm was proposed.The experimental results show that the proposed algorithm can quickly converge and accurately find the policy mapping even in the large-scale datasets.     

A behavior-driven reliability modeling method for complex smart systems

With recent advancements in Internet technologies and wireless communications, wireless sensor network (WSN)-based smart systems are gaining an enormous increase in use in various applications (eg, healthcare, smart home, smart manufacturing, smart power grids, and smart transportation). Due to the mission-critical or safety-critical nature of smart system applications, it is imperative that a smart system be reliable during its mission time. However, reliability analysis and design of smart systems are still open challenging research problems due to complicated dependencies existing in the WSN domain or the physical domain monitored by the WSN, and due to dependencies crossing the two domains. This paper proposes a new behavior-driven reliability modeling methodology for accurate and efficient reliability analysis of complex WSN-based smart systems, contributing toward their robust designs and operations. The suggested method can address effects of dependent behaviors affecting different domains of the smart system in a combinatorial manner. It also enables the use of efficient single-domain reliability methods to retain their efficiency. A case study of a smart home system is performed to demonstrate the application of the proposed method as well as its advantages in handling nonexponential time-to-failure distributions and in analyzing smart systems with complicated intradomain and cross-domain dependencies.     

基于多对多生成对抗网络的非对称跨域迁移行人再识别

无监督跨域迁移学习是行人再识别中一个非常重要的任务. 给定一个有标注的源域和一个没有标注的目标域, 无监督跨域迁移的关键点在于尽可能地把源域的知识迁移到目标域. 然而, 目前的跨域迁移方法忽略了域内各视角分布的差异性, 导致迁移效果不好. 针对这个缺陷, 本文提出了一个基于多视角的非对称跨域迁移学习的新问题. 为了实现这种非对称跨域迁移, 提出了一种基于多对多生成对抗网络(Many-to-many generative adversarial network, M2M-GAN)的迁移方法. 该方法嵌入了指定的源域视角标记和目标域视角标记作为引导信息, 并增加了视角分类器用于鉴别不同的视角分布, 从而使模型能自动针对不同的源域视角和目标域视角组合采取不同的迁移方式. 在行人再识别基准数据集Market1501、DukeMTMC-reID和MSMT17上, 实验验证了本文的方法能有效提升迁移效果, 达到更高的无监督跨域行人再识别准确率.     

基于联盟链的跨域认证方案

针对传统跨域认证易单点失效、过度依赖第三方等安全问题,提出了一种结合基于身份的密码（IBC,identity-based cryptography）体制与联盟链的跨域认证方案。通过设计包括实体层、代理层、区块链层、存储层在内的分层跨域认证架构,在跨域认证场景中引入联盟链,从而能够使两者较好地融合,增加了联盟链在跨域认证场景中的适应性。在存储层,设计摘要数据格式,将其存储于链上,摘要数据对应的完整数据存储于链下的星际文件系统,从而形成一种安全可靠的链上链下分布式存储方案,解决引入区块链后存在的链上存储受到限制的问题。提出一种基于永久自主权身份和临时身份的身份管理方案,解决结合IBC体制后身份难以注销和匿名身份难以监管的问题。在此基础上,设计完整的跨域全认证、重认证以及密钥协商协议以实现跨域认证流程。在安全性方面,使用 SVO 逻辑对认证协议进行分析,证明了跨域认证协议的安全性。通过仿真对计算负载性能、通信负载以及联盟链性能进行了测试与分析。分析表明,与相关方案相比,协议在满足安全性的同时,在服务端和用户端均有较好的计算负载表现。在通信效率上,相较于其他方案有不错的表现。通过联盟链工具对链上读写时延进行了测试,结果表明所提方案有良好的可用性。