Google has announced the general availability of AI Prediction Service, which is a critical component of its artificial intelligence platform. The service supports hosting the models trained in popular machine learning frameworks including XGBoost, Scikit-Learn, and TensorFlow.
Acting as the last stage of the machine learning pipeline, the AI Prediction Service hosts trained machine learning models in the cloud for inferring target values for new data. Trained models deployed in it are exposed as REST endpoints, which can be invoked from any HTTP-supportive standard client.
Based on the Google Kubernetes Engine (GKE) backend, the AI Platform Prediction service is designed for improved flexibility and reliability via new hardware options like Google Compute Engine machine types and NVIDIA GPUs. One of the best things about the AI Platform Prediction service is that it hides the complexity of provisioning, managing, and scaling the clusters despite being based on Google Kubernetes Engine. Instead of managing the infrastructure, data scientists and engineers can focus on business problems.
With the general availability, AI Prediction service supports Scikit-Learn and XGBoost models on high-CPU and high-memory machine types. The service, behind the scenes, automatically expands and shrinks the infrastructure depending on the requests and traffic. This service is closely synchronized with Google Cloud Console and Stackdriver for tracking and visualizing resource metrics.
Customers can now choose to deploy machine learning models in specific regions via the AI Prediction Service. Models that get deployed on the regional endpoints stay within the specified region to provide data sovereignty and locality to customers.
In the last few years, Google has been investing heavily in the AI Platform as a Service (PaaS) offering. Its consolidated and augmented various services have now evolved into an end-to-end platform that supports data preparation, transformation, training, deployment, model management, model management wth tight integration with GKE and Kubeflow.
Need help with Google Cloud integration, development, or consultancy? Reach us at +1 (415) 830-3899 or at info@cloudanalogy.com now.
