机器学习模型部署_9月版部署机器学习模型_机器学习

机器学习模型部署

每月版 (MONTHLY EDITION)

Often, the last step of a Data Science task is deployment. Let’s say you’re working at a big corporation. You’re building a project for a customer of the corporation and you’ve created a model that performs well. Unfortunately, the model you’ve created will only be able to be used by the customer if the customer has the code you’ve written, the environment you’ve created, and the machines you’ve been working on.

通常，数据科学任务的最后一步是部署。假设您在一家大公司工作。您正在为公司的客户构建项目，并且已经创建了一个运行良好的模型。不幸的是，只有当客户拥有您编写的代码，所创建的环境以及正在使用的机器时，客户才能使用您创建的模型。

HOWEVER, if you deploy your model into production, the only thing the customer will need is…the product. In other words, a machine learning model will provide real value when it is available to the users that it has been created for. Your model is only a proof of concept (PoC) until it is put into production, then it becomes a deliverable.

但是，如果您将模型部署到生产中，那么客户唯一需要的就是产品。换句话说， 当机器学习模型可供创建的用户使用时，它将提供真正的价值 。您的模型只是概念证明(PoC)，直到投入生产，然后才能交付使用。

There are many ways to deploy a machine learning model. The basic idea of deployment involves allowing an end-user to utilize your model. The product needs to be customized to the end user’s needs since they will be the ones who will use it. Deployment is a crucial step because it allows others to use the machine learning model that was built.

有很多方法可以部署机器学习模型。 部署的基本思想涉及允许最终用户使用您的模型 。产品需要根据最终用户的需求进行定制，因为他们将是使用产品的人。部署是至关重要的一步，因为它允许其他人使用已构建的机器学习模型。

Choosing how to deploy your model into production can be difficult and you’ll need to evaluate what the end-users want and need. Perhaps your model needs to work in real time. Maybe it needs to be used to make many predictions at a time. You might need a particular architecture, etc. There can be many many requirements for a product, and more importantly, it will need to work on all use-cases, which is why debugging your model is essential.

选择如何将模型部署到生产环境可能很困难，您需要评估最终用户的需求。也许您的模型需要实时工作。也许需要一次使用它进行许多预测。您可能需要特定的体系结构等。产品可能有很多需求，更重要的是，它将需要在所有用例上工作，这就是调试模型至关重要的原因。

Michael Armanious, Editor at Towards Data Science.

《迈向数据科学》(Towards Data Science)编辑Michael Armanious 。