Hands-On Artificial Intelligence for Beginners
上QQ阅读APP看书,第一时间看更新

GCP Cloud ML Engine

Google Cloud Platform's Cloud ML Engine is Google's equivalent to AWS SageMaker. As a managed PaaS, Cloud ML handles the training and deployment processes for machine learning algorithms. If you're thinking - what about a basic compute service like EC2 on AWS? GCP has that as well. Compute Engine is GCP's answer to Amazon EC2; it provides basic, scalable cloud compute services. While we could use Compute Engine to setup AI platforms, GCP has made it extremely simple for us to build with Cloud ML Engine and as such, we will note be covering the basic Compute Engine. 

Let's dive into the details. Cloud ML engine allows you to: 

  • Train scikit-learn and TensorFlow models both locally for testing and in the cloud
  • Create retrainable machine learning models that are stored in the cloud
  • Easily deploy trained models to production

Cloud ML jobs are setup through the terminal. We'll work on running these training jobs in the coming chapters as we start to work with various ANN models.