Top 6 Notable Updates to Amazon AWS Sagemaker Service

In the past year, exponential progress has been seen in Machine Learning and customers are using the Amazon Sagemaker service. AWS Sagemaker is helping them to perform various tasks such as finding fraud, tune engines and predict pitches as well. About 100 new features have been added to the original product as a result of the constant customer feedback the company has been receiving from the developers.

Overall AWS Sagemaker received updates broadly categorized as,

• Workflows
• New Algorithms
• Search additions
• Git Integration
• Apache Airflow
• Standards and accredeitions 

Interest in more details, then read on,

Sagemaker Workflows

Amazon has introduced workflows which will help with the automation, orchestration, and collaboration of features, thus making it easier for developers to build, share as well as manage the ML workflows. Since machine learning consists of various inputs from different types of data sources, it often requires multiple iterations and experimentations with features. Just leads to developers wanting to share their progress with each other and use the successful parts and reuse the old ones. Sagemaker Workflows intends to achieve the same with the introduction of these new capabilities.

Sagemaker Search additions

Since the development of an ML model requires continuous experimentation with newly developed algorithms and models, Sagemaker Search will help data scientists and developers alike enabling them to quickly find and evaluate the model that fits well from the potential list of thousands of models right from the AWS console.

Git Integration

The ability to integrate with Git and visualization to Amazon Sagemaker will further enhance the process of sharing ideas, tasks and the overall progress of the machine learning process.

Customers are now able to link their GitHub and AWS CodeCommit with Sagemaker notebooks as well as a clone, make public or private repositories and store the information securely using IM, LDAP, and the AWS Secrets Manager. The new open sourced notebook app allows developers to review the branches, merges, and versions directly via Sagemaker.

Apache Airflow and Step Functions

A coordinated number of sequences have to be followed which involve multiple steps to complete workflow in machine learning. For example, you may want to perform a query in Amazon Athena or aggregate and prepare data in AWS Glue, before training a model in SageMaker, and deploying it to production. Automating these steps and orchestrating them across multiple services helps build reusable, reproducible ML workflows which can be shared between engineers and scientists.

New Algorithms and Frameworks

The new Sagemaker features provide three functions for testing algorithms for training models: by bringing in their own custom container, using inbuilt Sagemaker algorithms, or running MXNet, TensorFlow, Pytorch, and Chainer algorithms in just 20 lines of code.

New Compliance Standards and Accreditation

SageMaker has been added to the existing System and Organizational Controls (SOC) Level 1, 2 and 3 audits. These SOC reports are now available directly in the AWS Console, with the level 3 report available as a PDF.

The recently introduced features offer new capabilities, algorithms, and accreditation which will result in bringing in more machine learning workloads for more developers. Amazon said that it has been focusing exclusively on the customer feedback and introducing tools which will directly benefit them.