Google announced the Cloud AI Platform, which provides users with a way to deploy robust, repeatable machine learning pipelines and monitoring, auditing, version tracking, and reproducibility. With Cloud AI Platform, Google can help organizations adopt the Machine Learning Operations practice, also known as MLOps – a term used to apply DevOps practices to help users automate, manage, and audit ML workflows. These practices typically involve data preparation and analysis, training, assessment, deployment, and more. Artificial Intelligence is a revolution, and it’s time you should upgrade your skills by registering for artificial intelligence training and receive your very own AI certification.
Knowledge of Blog
- How Does It Run?
- Inclusions
- Planned Features
- Price Details
- Conclusion
Google Cloud Platform (GCP) declared a new beta offering to simplify the deployment of complicated machine learning workflows that often have many moving, interdependent parts. Let’s know more.
How Does It Run?
AI Platform, Pipelines has two main components:
- The infrastructure for deploying and operating structured AI workflows integrated with Google Cloud Platform services
- The pipeline tools for debugging, building, and sharing pipelines and components.
The service runs on an automatically created Google Kubernetes cluster as part of the installation process and is accessible via the Cloud AI Platform’s dashboard. AI developers specify a pipeline using the Kube Flow Pipelines (KFP) software development kit (SDK) with AI Platform Pipelines, or by customizing the TensorFlow Extended (TFX) pipeline template with the TFX SDK.
AI Pipelines operates the pipeline using the open-source Argo workflow engine and has additional microservices to record metadata, handle IO components, and schedule pipeline runs. As individual isolated pods in a cluster, pipeline steps are executed, and each component can leverage Google Cloud services such as AI Platform Training and Prediction, BigQuery, Dataflow, and others. Meanwhile, the pipelines may contain steps that perform the cluster’s graphics card and tensor processing unit’s computation, directly leveraging features such as autoscaling and auto-provisioning of nodes.
Inclusions
AI Platform Pipeline includes automatic metadata tracking using ML Metadata, a library for recording and recovering metadata associated with developers of machine learning and data scientists’ workflows. Automatic metadata tracking logs the artifacts applied in each step of the pipeline, the pipeline’s parameters, and the linkage through the artifacts input/output, as well as the steps of the pipeline that created and consumed them.
Additionally, AI Platform Pipelines supports pipeline versioning, enabling developers to upload and group multiple versions of the same pipeline in the UI and automatic artifact and lineage tracking.
Native artifact tracking enables stuff like model tracking, data statistics, model evaluation metrics, etc. And lineage tracking displays your model history and versions, data, and more.
The product equips organizations with:
- Push-button installation via the Google Cloud Console
- Enterprise specialties for running ML workloads, including pipeline versioning, automatic metadata tracking of artifacts and executions, Cloud Logging, visualization tools, and more
- Seamless amalgamation with Google Cloud managed services like BigQuery, Dataflow, AI Platform Training and Serving, Cloud Functions, and many others
- Several prebuilt pipeline components (pipeline steps) for ML workflows, with easy construction of your custom components
AI Platform makes it simple for machine learning developers, data scientists, and data engineers to quickly and cost-effectively take their ML projects from the idea to the production and deployment. AI Platform’s integrated toolchain helps build and run your machine learning applications, from data engineering to “no lock-in” flexibility.
You can store your data using Cloud Storage or BigQuery. Apply the built-in data labeling service to label your training data for images, videos, audio, and text by applying classification, object detection, entity extraction, etc. You can also import the etiquette data into AutoML and directly train a model.
Using the Deep Learning VM Image, you can build your ML applications on GCP using a managed Jupyter Notebook service that provides fully configured environments for various ML frameworks. You can then use AI Platform Training and Prediction services to train your models and deploy them to GCP production in a serverless environment or do so on-site using Kubeflow’s training and predictive microservices.
Planned Features
AI Platform Pipelines will gain multi-user isolation shortly, allowing individuals accessing the Pipelines Cluster Control to access their pipelines and other resources. Other features to come to include workload identity to support transparent access to Google Cloud Services; UI-based setup of backend data storage off-cluster, including metadata, server data, job history, and metrics; more straightforward cluster upgrades; and more templates to workflow authoring.
Going ahead, planned new features comprise:
- Multi-user isolation, so that any person accessing the Pipelines cluster can regulate who can access their pipelines and other resources
- Workload identity, to maintain transparent access to GCP services
- Light, UI-based setup of off-cluster storage of backend data—including metadata, server data, job history, and metrics—for larger-scale deployments and so that it can persist after cluster shutdown
- Easy cluster upgrades
- More templates for authoring ML workflows
Price Details
You can use Kube Flow, AI Hub, and notebooks at no charge. Google’s pricing calculator can also be used to estimate the costs of running your workload.
The Google Cloud Platform provides $300 credits for free on the first Sign up, and users get $300 to spend on Google Cloud Platform over the next 12 months.
There’s also no auto-charge after the free trial ends. They ask you for your credit card to make sure you are not a robot. You won’t be charged unless you manually upgrade to a paid account.
Conclusion
Google’s Cloud AI Platform surely simplifies machine learning development. Easy-to-use artificial intelligence and machine learning capacities are embedded in Google’s core solutions and infrastructure, making them accessible and easily deployed. Complete AI and ML toolsets give you the support you need to recognize and address issues, drive new business insights, and better serve your customers. Enterprises over industries are taking benefit of Google technology advances—from embedded AI and ML to Digital Assistant, Search, Maps, Voice, and more—that are seamlessly integrated into Google Cloud. Become an artificial intelligence developer and increase your chances of being hired!