Workloads including artificial intelligence (AI) and machine learning (ML) can be executed everywhere, including on-premises, at the edge, embedded in hardware, and in the cloud.
Amazon Web Services (AWS), which is providing an expanding range of services, is betting that businesses would frequently select the cloud. Today at the AWS re: invent 2022 event in Las Vegas, the firm unveiled key components of its AI/ML strategy as well as a bewildering array of new features and services that will aid businesses in using the cloud for data science.
The SageMaker suite of products is the keystone of the AWS AI/ML portfolio. Swami Sivasubramanian, VP database, analytics, and ML at AWS, stated that SageMaker enables enterprises to build, train, and deploy ML models for almost any use case and provides tools for every phase of ML development in a keynote talk at AWS re:invent.
According to Sivasubramanian, “tens of thousands of customers are utilizing SageMaker ML models to create more than a trillion predictions a month. By leveraging that data to create ML models, their customers are using SageMaker to solve complex challenges ranging from expediting drug development to optimizing driving routes for rideshare apps.
SageMaker now supports geospatial ML
SageMaker’s feature set is currently being enhanced with enhanced geospatial ML capabilities.
According to Sivasubramanian, geospatial data can be used for a wide range of applications. It can be used to help optimize an agricultural harvest yield, aid in planning for sustainable urban development, and identify a new location or region for a business to open.
Working with numerous data sources and vendors is necessary to obtain high-quality geospatial data for ML model training, according to him. These data sets are frequently enormous and unstructured, so time-consuming data preparation is required before you can write a single line of code to create your machine learning models.
With the addition of geospatial capabilities in SageMaker, AWS hopes to simplify the actual development and deployment of models for businesses. According to Sivasubramanian, the new feature would allow users to quickly and easily access geospatial data in SageMaker from a variety of data sources.
SageMaker has recently included geospatial data preparation tools to aid users in processing and enhancing huge datasets. SageMaker now has integrated visualization features that let users explore model predictions on an interactive map while analyzing data using 3D accelerated graphics.
For many popular geospatial use cases, Sivasubramanian continued, SageMaker now also includes built-in pretrained neural nets.
ML Governance is strengthened
Collaboration across groups is becoming more and more important as firms integrate ML into various processes.
Another area where AWS is seeking to assist its users with new capabilities in the Amazon SageMaker ML Governance service is in the development of the permissions and governance rules that enable model sharing. SageMaker Role Manager, Model Cards, and Model Dashboard are some of the new services.
SageMaker Role Manager, according to Sivasubramanian, assists businesses in defining important permissions for users through automated policy generation tools. The Model Cards service aims to create a centralized authoritative repository for ML model documentation. The new Model Dashboard now gives enterprises visibility and unified monitoring of ML model performance.
These are quite powerful governance skills that will assist you in responsibly building ML governance, Sivasubramanian added.
