ID 原文 译文
17565 其次，以负载均衡为资源协调原则，与VNF可靠性联合优化，最终使用深度强化学习得到服务功能链部署策略。 Secondly, taking load balancing as theresource coordination principle, joint optimization the VNF reliability is jointly optimized. Finally, the deepreinforcement learning is used to get the service function chain deployment strategy.
17566 另外，提出了基于重要度的节点备份和链路备份策略，用于应对部署过程中VNF/链路可靠性难以满足的情况。 In addition, node backupand link backup strategies based on importance are proposed to deal with situations where VNF/link reliabilityis difficult to meet during deployment.
17567 仿真结果表明，该文的可靠部署算法在保证可靠性需求的基础上能够有效减少SFC失效损失，同时使虚拟网络更加稳定可靠。 Simulation results show that the reliable deployment algorithm in thispaper can effectively reduce the failure SFC loss on the basis of ensuring the reliability requirements, and at thesame time make the virtual network more stable and reliable.
17568 尽管由于丢弃维度将3维(3D)形状投影到2维(2D)视图看似是不可逆的，但是从可视化到计算机辅助几何设计，各个垂直行业对3维重建技术的兴趣正迅速增长。 While projecting 3D shapes to 2D images is irreversible due to the abandoned dimension amid the projection process, there are rapidly growing interests across various vertical industries for 3D reconstruction techniques, from visualization purposes to computer aided geometric design.
17569 传统基于物体深度图或者RGB图的3维重建算法虽然可以在一些方面达到令人满意的效果，但是它们仍然面临若干问题：(1)粗鲁的学习2D视图与3D形状之间的映射； The traditional 3D reconstructionapproaches based on depth map or RGB image can synthesize visually satisfactory 3D objects, while theygenerally suffer from several problems:
17570 (2)无法解决物体不同视角下外观差异所带来的的影响；(3)要求物体多个观察视角下的图像。 (1)The 2D to 3D learning strategy is brutal-force; (2)Unable to solve theeffects of differences in appearance from different viewpoints of objects; (3)Multiple images from distinctlydifferent viewpoints are required.
17571 该文提出一个端到端的视图感知3维(VA3D)重建网络解决了上述问题。 In this paper, an end-to-end View-Aware 3D (VA3D) reconstruction network is proposed to address the above problems.
17572 具体而言，VA3D包含多邻近视图合成子网络和3D重建子网络。 In particular, the VA3D includes a multi-neighbor-view synthesissub-network and a 3D reconstruction sub-network.
17573 多邻近视图合成子网络基于物体源视图生成多个邻近视角图像，且引入自适应融合模块解决了视角转换过程中出现的模糊或扭曲等问题。 The multi-neighbor-view synthesis sub-network generatesmultiple neighboring viewpoint images based on the object source view, while the adaptive fusional module isadded to resolve the blurry and distortion issues in viewpoint translation.
17574 3D重建子网络使用循环神经网络从合成的多视图序列中恢复物体3D形状。 The 3D reconstruction sub-networkintroduces a recurrent neural network to recover the object 3D shape from multi-view sequence.