高级检索

    视频云网平台中智能算法版权管理方法

    Algorithmic Intelligence Right Management Method in Video Cloud-Network Platform

    • 摘要: 视频云网平台中涵盖了大量智能算法,如何对其进行高效管理,从而支持应用服务的快速部署与更新是一个重要的科学问题. 然而,传统的智能算法与云端资源具有绑定规则,不同应用服务商之间的智能算法缺乏统一的调用机制,导致它们无法快速整合和有效利用. 为了解决此难题,建立“服务—算法—资源”动态互联服务体系,有效解决算法快速迭代、应用需求时变与智能算法版权固化管理的矛盾. 在动态互联服务过程中,传统的、面向固定内容的买断式数字版权管理已经无法为细粒度权限管理提供高效服务. 为此,提出智能算法版权管理系统(algorithmic intelligence right management,AIRM),通过设计版权资源服务化方法与流动性算力网络结构,构建视频云网平台中“共享式”智能算法版权管理方法.在中国电信视频分析平台授权管理模块中的实际部署结果表明,所设计方法可以将算法并发服务能力提高19.9倍,将算法版权响应时间降低18.36%.

       

      Abstract: Video cloud-network platform contains a huge amount of intelligent algorithms, and it is an important scientific problem to efficiently manage video cloud-network platform so as to support the rapid deployment and update of application services. However, the traditional intelligent algorithms are forcibly bounded to cloud resource, and there is no unified invocation mechanism for intelligent algorithms among different service providers, which is difficult for fast integration and effective utilization. In order to solve this problem, we propose the “service-algorithm-resource” dynamic interconnection service system, which can effectively solve the contradiction among rapid iteration of algorithm, dynamic application demand and fixed management of intelligent algorithms. In the process of dynamic interconnection services, the traditional and fixed content-oriented buyout digital rights management can no longer provide efficient services for fine-grained rights management. To this end, we propose an algorithmic intelligence right management (AIRM) system, and build “sharing-mode” intelligent algorithm right management method on the video cloud-network platform through right resource servitization method and liquidity arithmetic network structure. The actual deployment results in the China telecom video analysis platform authorization management module show that the designed method can increase the parallel algorithm service capacity by 19.9 times, and decrease the right response time by 18.36%.

       

    /

    返回文章
    返回