AWS Releases New Capabilities for Amazon SageMaker

At its re: Invent 2021, AWS has announced exciting capabilities for its Amazon SageMaker capabilities.

At Amazon’ AWS re: Invent 2021, AWS announced new capabilities for its machine learning service platform SageMaker, making machine learning even more powerful, more accessible, and at a reasonable sum. The announcements bring robust capabilities that include a no-code ecosystem for developing precise data labeling using highly skilled annotators. This will provide a universal experience for users while enhancing collaboration across domains, providing a compiler for machine learning training making code much more efficient. Not only that, the SageMaker experience also provides automatic compute instance selection machine learning inference and serverless compute for machine learning inference.

With customers aiming to scale their machine learning training model, AWS has continued to invest in expanding the service’s capabilities to deliver over 60 new Amazon SageMaker functionality in 2020. The announcement of the new capabilities is based on these advancements making it even easier to prepare as well as gather data for machine learning, training models at a rapid pace, optimize the type as well as amount of computing required for inference.

• No-code machine learning predictions

Broadening access to machine learning by giving business analysts a visual interface, Amazon SageMaker Canvas allows them to develop accurate machine learning predictions on their own without needing any machine learning experience or having to write code. Users can point Amazon SageMaker Canvas to their data stores as well as the Amazon SageMaker that deliver visual tools to help them intuitively prepare and analyze data. It allows business analysts to review as well as evaluate models in the Amazon SageMaker Canvas console for accuracy as well as efficacy of their usage. It lets users export their models to Amazon SageMaker Studio. This allows them to share their data with data scientists to validate and further refine their models.

• Amazon Sage Ground Truth Plus

A fully-managed data labeling service, Amazon SageMaker Ground Truth Plus utilizes an expert staff with built-in annotation workflows. This allows them to transport high-quality data for training machine learning models at a lower cost with no coding required. The new capability makes it seamless for customers to generate labeled data by using human annotators via Amazon Mechanical Turk, private workforce, or third-party vendors. This allows it to meet the customers’ data security as well as qualifications to meet customers’ data security, compliance and privacy requirements for accurate data labeling.

Also Read: Three FinTech Trends to Watch in 2022

• Amazon SageMaker Studio universal notebooks

Being a universal notebook for Amazon SageMaker Studio, it provides a single, integrated platform to perform data engineering, machine learning and analytics. This lets the users interactively transform, access as well as analyze a broad range of data for multiple purposes within a universal notebook.

• Training Compiler for Machine Learning Models

The latest machine learning model compiler optimizes code to utilize computing resources effectively while significantly reducing the time needed to train models. Integrated with the versions TensorFlow and PyTorch in Amazon SageMaker, it has been optimized to run with utmost efficiency in the cloud environment, allowing data scientists to utilize their preferred frameworks to train machine learning models.

• Inference Recommender automatic instance selection

It assists customers in choosing the best computer instance as well as configuration to power a specific machine learning model. For big machine learning models, commonly referred to as natural language processing or computer vision, choosing a compute instance having the best price for performance is a complicated, iterative process that can take weeks of experimentation. This removes the guesswork for customers as well as complexity to determine where to run a specific model while reducing the time to deploy from weeks to hours. 

Check Out The New Enterprisetalk Podcast. For more such updates follow us on Google News Enterprisetalk News.