一种用于增量式学习云系统的训练、调度方法及相关设备技术方案

一种用于增量式学习云系统的训练、调度方法及相关设备，涉及数据处理领域。该训练方法可包括训练云接收识别云发送模型训练请求；根据所述识别信息和所述识别模型的类型生成对应的训练任务(302)；通过所述识别信息计算所述训练任务的优先数(303)，所述训练任务的优先数对应所述训练任务的执行优先级别；根据所述优先数为所述训练任务分配训练资源，并按所述执行优先级别执行对应的训练任务(304)。通过计算训练任务的优先数来确定训练任务的优先级别，根据计算出的优先数对训练云中得训练任务进行调度，使得训练资源能够被多个训练任务合理共享，提高训练效率。

Training, scheduling method and related equipment for incremental learning cloud system

A training, scheduling method and related equipment for incremental learning cloud system relates to the field of data processing. The training method may include training cloud receiving and identifying cloud transmission model training requests, generating corresponding training tasks (302) according to the identification information and the type of the identification model, and calculating the priority number (303) of the training task by the identification information, and the priority number of the training task corresponds to the execution of the training task. First level; according to the priority number, the training resources are allocated for the training tasks, and the corresponding training tasks are executed according to the execution priority level (304). The priority of the training task is determined by calculating the priority number of the training task, and the training task is scheduled according to the calculated priority number, so that the training resources can be shared reasonably by several training tasks and improve the training efficiency.

【技术实现步骤摘要】
【国外来华专利技术】PCT国内申请，说明书已公开。

【技术保护点】
PCT国内申请，权利要求书已公开。

【技术特征摘要】
【国外来华专利技术】PCT国内申请，...

【专利技术属性】
技术研发人员：邵云峰，姚骏，薛希俊，
申请(专利权)人：华为技术有限公司，
类型：发明
国别省市：广东,44