您现在访问的是微软AZURE全球版技术文档网站,若需要访问由世纪互联运营的MICROSOFT AZURE中国区技术文档网站,请访问 https://docs.azure.cn.

什么是 Azure 机器学习设计器(预览版)?What is Azure Machine Learning designer (preview)?

应用于:否基本版是企业版            (升级到企业版APPLIES TO: noBasic edition yesEnterprise edition                       (Upgrade to Enterprise)

使用 Azure 机器学习设计器,能够在交互式画布上直观地连接数据集模块,以创建机器学习模型。Azure Machine Learning designer lets you visually connect datasets and modules on an interactive canvas to create machine learning models. 要了解如何开始与设计师合作,请参阅教程:与设计师一起预测汽车价格To learn how to get started with the designer, see Tutorial: Predict automobile price with the designer

Azure 机器学习设计器示例

设计器使用 Azure 机器学习工作区来整理共享资源,例如:The designer uses your Azure Machine Learning workspace to organize shared resources such as:

模型训练和部署Model training and deployment

设计器提供用于生成、测试和部署机器学习模型的可视化画布。The designer gives you a visual canvas to build, test, and deploy machine learning models. 使用设计器,可以:With the designer you can:

  • 数据集模块拖放至画布上。Drag-and-drop datasets and modules onto the canvas.
  • 将模块连接在一起以创建管道草稿Connect the modules together to create a pipeline draft.
  • 使用 Azure 机器学习工作区中的计算资源提交管道运行Submit a pipeline run using the compute resources in your Azure Machine Learning workspace.
  • 将训练管道转换为推理管道********。Convert your training pipelines to inference pipelines.
  • 将管道发布到 REST 管道终结点,以提交具有不同参数和数据集的新管道运行****。Publish your pipelines to a REST pipeline endpoint to submit new pipeline runs with different parameters and datasets.
    • 发布训练管道,在更改参数和数据集时重用单个管道训练多个模型****。Publish a training pipeline to reuse a single pipeline to train multiple models while changing parameters and datasets.
    • 发布批量推理管道,通过使用以前训练的模型针对新数据进行预测****。Publish a batch inference pipeline to make predictions on new data by using a previously trained model.
  • 将实时推理管道部署到实时终结点,以便针对新数据进行实时预测****。Deploy a real-time inference pipeline to a real-time endpoint to make predictions on new data in real time.

设计器中的训练、批量推理和实时推理的工作流示意图

管道Pipeline

管道包含连接在一起的数据集和分析模块。A pipeline consists of datasets and analytical modules, which you connect together. 管道有许多用途,可以创建管道来训练单个模型或多个模型。Pipelines have many uses: you can make a pipeline that trains a single model, or one that trains multiple models. 可以创建管道来进行实时预测或批量预测,或者仅用于清理数据。You can create a pipeline that makes predictions in real time or in batch, or make a pipeline that only cleans data. 借助管道,可以重复使用工作成果和整理项目。Pipelines let you reuse your work and organize your projects.

管道草稿Pipeline draft

在设计器中编辑管道时,你的进度会保存为管道草稿****。As you edit a pipeline in the designer, your progress is saved as a pipeline draft. 可以通过添加或删除模块、配置计算目标、创建参数等方式随时编辑管道草案。You can edit a pipeline draft at any point by adding or removing modules, configuring compute targets, creating parameters, and so on.

有效管道具有以下特征:A valid pipeline has these characteristics:

  • 数据集只能连接到模块。Datasets can only connect to modules.
  • 模块只能连接到数据集或其他模块。Modules can only connect to either datasets or other modules.
  • 模块的所有输入端口必须与数据流建立某种连接。All input ports for modules must have some connection to the data flow.
  • 必须设置每个模块的所有必需参数。All required parameters for each module must be set.

若准备好运行管道草稿,请提交管道运行。When you're ready to run your pipeline draft, you submit a pipeline run.

管道运行Pipeline run

每次运行管道时,管道及其结果的配置都作为管道运行存储在工作区中****。Each time you run a pipeline, the configuration of the pipeline and its results are stored in your workspace as a pipeline run. 出于故障排除或审核目的,可以返回任何管道运行以对其进行检查。You can go back to any pipeline run to inspect it for troubleshooting or auditing purposes. 克隆管道运行,可创建新的管道草稿以供编辑****。Clone a pipeline run to create a new pipeline draft for you to edit.

管道运行被分入试验以整理运行历史记录。Pipeline runs are grouped into experiments to organize run history. 可以为每个管道运行设置试验。You can set the experiment for every pipeline run.

数据集Datasets

使用机器学习数据集可以轻松地访问和处理数据。A machine learning dataset makes it easy to access and work with your data. 此设计器中包含很多示例数据集供你进行试验。A number of sample datasets are included in the designer for you to experiment with. 你可以根据需要注册更多数据集。You can register more datasets as you need them.

模块Module

模块是可对数据执行的算法。A module is an algorithm that you can perform on your data. 设计器有许多模块,包括数据引入函数、训练、评分和验证过程。The designer has a number of modules ranging from data ingress functions to training, scoring, and validation processes.

模块可能提供一组参数用于配置模块的内部算法。A module may have a set of parameters that you can use to configure the module's internal algorithms. 在画布上选择模块时,模块的参数会显示在画布右侧的“属性”窗格中。When you select a module on the canvas, the module's parameters are displayed in the Properties pane to the right of the canvas. 可以在该窗格中修改参数来调整模型。You can modify the parameters in that pane to tune your model. 可以在设计器中设置各个模块的计算资源。You can set the compute resources for individual modules in the designer.

模块属性

在浏览可用的机器学习算法库时如需帮助,请参阅算法和模块参考概述For some help navigating through the library of machine learning algorithms available, see Algorithm & module reference overview

计算资源Compute resources

使用工作区中的计算资源来运行管道,并将已部署的模型作为实时终结点或管道终结点托管(用于批量推理)。Use compute resources from your workspace to run your pipeline and host your deployed models as real-time endpoints or pipeline endpoints (for batch inference). 支持的计算目标为:The supported compute targets are:

计算目标Compute target 培训Training 部署Deployment
Azure 机器学习计算Azure Machine Learning compute
Azure Kubernetes 服务Azure Kubernetes Service

计算目标会附加至 Azure 机器学习工作区Compute targets are attached to your Azure Machine Learning workspace. 可在 Azure 机器学习工作室(经典)中管理工作区中的计算目标。You manage your compute targets in your workspace in Azure Machine Learning Studio (classic).

部署Deploy

若要执行实时推理,必须将管道部署为实时终结点To perform real-time inferencing, you must deploy a pipeline as a real-time endpoint. 实时终结点在外部应用程序和评分模型之间创建接口。The real-time endpoint creates an interface between an external application and your scoring model. 对实时端点的调用会将预测结果实时返回至应用程序。A call to a real-time endpoint returns prediction results to the application in real time. 若要调用实时终结点,请传递部署终结点时创建的 API 密钥。To make a call to a real-time endpoint, you pass the API key that was created when you deployed the endpoint. 该终结点基于 REST,这是一种流行的 Web 编程项目的体系结构。The endpoint is based on REST, a popular architecture choice for web programming projects.

必须将实时终结点部署到 Azure Kubernetes 服务群集。Real-time endpoints must be deployed to an Azure Kubernetes Service cluster.

要了解如何部署模型,请参阅教程:使用设计器部署机器学习模型To learn how to deploy your model, see Tutorial: Deploy a machine learning model with the designer.

发布Publish

还可以将管道发布到管道终结点****。You can also publish a pipeline to a pipeline endpoint. 与实时终结点类似,借助管道终结点,可以使用 REST 调用从外部应用程序提交新的管道运行。Similar to a real-time endpoint, a pipeline endpoint lets you submit new pipeline runs from external applications using REST calls. 但是不能使用管道终结点实时发送或接收数据。However, you cannot send or receive data in real-time using a pipeline endpoint.

已发布的管道是灵活的,它们可用于训练或重新训练模型、执行批量推断、处理新数据等。Published pipelines are flexible, they can be used to train or retrain models, perform batch inferencing, process new data, and much more. 可以将多个管道发布到单个管道终结点,并指定要运行的管道版本。You can publish multiple pipelines to a single pipeline endpoint and specify which pipeline version to run.

已发布的管道在每个模块的管道草稿中定义的计算资源上运行。A published pipeline runs on the compute resources you define in the pipeline draft for each module.

设计器创建与 SDK 相同的 PublishedPipeline 对象。The designer creates the same PublishedPipeline object as the SDK.

从可视化界面移动到设计器Moving from the visual interface to the designer

可视化界面(预览)已更新,现在是 Azure 机器学习设计器(预览)。The visual interface (preview) has been updated and is now Azure Machine Learning designer (preview). 设计器经过重建,可使用基于管道的后端,与 Azure 机器学习的其他功能完全集成。The designer has been rearchitected to use a pipeline-based backend that fully integrates with the other features of Azure Machine Learning.

通过这些更新,可视化界面的一些概念和术语已产生变化或更改了名称。As a result of these updates, some concepts and terms for the visual interface have been changed or renamed. 请参阅下表,了解最关键的概念更改。See the table below for the most important conceptual changes.

设计器中的概念Concept in the designer 旧版(在可视化界面中)Previously in the visual interface
管道草稿Pipeline draft 试验Experiment
实时终结点Real-time endpoint Web 服务Web service

迁移到设计器Migrating to the designer

可将现有可视化界面试验和 Web 服务转换为设计器中的管道和实时终结点。You can convert existing visual interface experiments and web services to pipelines and real-time endpoints in the designer. 要迁移可视化界面资产,请执行以下步骤:Use the following steps to migrate your visual interface assets:

  1. 登录到 Azure 机器学习工作室Sign in to Azure Machine Learning studio.

  2. 将工作区升级到企业版。Upgrade your workspace to Enterprise edition.

    升级后,所有视觉对象界面试验都将转换为设计器中的管道草稿。After upgrading, all of your visual interface experiments will convert to pipeline drafts in the designer.

    备注

    不需升级到企业版本即可将视觉对象界面 Web 服务转换为实时终结点。You don't need to upgrade to the Enterprise edition to convert visual interface web services to real-time endpoints.

  3. 请在工作区的设计器部分查看管道草稿列表。Go to the designer section of the workspace to view your list of pipeline drafts.

    转换后的 Web 服务可以通过导航到端点 > 实时终结点找到。Converted web services can be found by navigating to Endpoints > Real-time endpoints.

  4. 选择要打开的管道草案。Select a pipeline draft to open it.

    如果在转换过程中出现错误,则会显示错误消息,其中包含用于解决此问题的说明。If there was an error during the conversion process, an error message will appear with instructions to resolve the issue.

已知问题Known issues

下面是需要手动解决的已知迁移问题:Below are known migration issues that need to be addressed manually:

  • “导入数据”或“导出数据”模块********Import Data or Export Data modules

    如果在试验中有“导入数据”或“导出数据”模块********,则需更新数据源才能使用数据存储。If you have an Import Data or Export Data module in the experiment, you need to update the data source to use a datastores. 若要了解如何创建数据存储,请参阅如何访问 Azure 存储服务中的数据To learn how to create a datastore, see How to Access Data in Azure storage services. 你的云存储帐户信息已添加到“导入数据”或“导出数据”模块,******** 方便你使用。Your cloud storage account information have been added in the comments of the Import Data or Export Data module for your convenience.

后续步骤Next steps