At this year’s AWS re:Invent, AWS announced 13 new machine learning services and capabilities. The company introduced SageMaker features to help developers build, train, and deploy machine learning modules. They also revealed services, framework enhancements, as well as a custom chip to speed up machine learning training and inference. Additionally, AWS announced new AI services that extract text from almost any document, read medical information, and provide customized personalization, recommendations, and forecasts. The final announcement involved AWS DeepRacer, a new 1/18th scale autonomous model race car driven by reinforcement learning.
New Infrastructure for Faster Training
Machine learning models typically train within an algorithm that finds patterns in data. This requires significant computing power, often inaccessible on-prem. AWS hopes to eliminate speed bumps in machine learning with a stronger infrastructure.
Amazon Elastic Compute Cloud (EC2) GPU Instances
New P3dn.24xl instances are the most powerful machine learning training processors available in the cloud, allowing developers to train models with more data in less time.
AWS-Optimized TensorFlow framework
Developers using TensorFlow have found difficulty scaling across multiple GPUs, leading to low utilization and longer training times. Thus, AWS worked to resolve this problem by allowing TensorFlow to scale across GPUs.
Inference accounts for most of the cost and complexity in running machine learning in production. In fact, for every dollar spent on training, nine are spent on inference. Amazon Elastic Inference allows developers to decrease inference costs with up to 75 percent savings when compared to the cost of using a dedicated GPU instance.
AWS Inferentia provides hundreds of teraflops per chip and thousands of teraflops per Amazon EC2 instance for multiple frameworks and data types.
New Amazon SageMaker capabilities
Amazon SageMaker is a fully managed service to ease the machine learning process so developers can build, train, and deploy machine learning modules.
Amazon SageMaker Ground Truth makes it easier for developers to label their data using human annotators through Mechanical Turk, third-party vendors, or their own employees. Amazon SageMaker Ground Truth learns from these annotations in real time and can automatically apply labels to much of the remaining dataset, reducing the need for human review.
The new AWS Marketplace for Machine Learning includes over 150 algorithms and models (with more coming every day) that deploy directly to Amazon SageMaker.
Amazon SageMaker RL, the cloud’s first managed reinforcement learning service, allows any developer to build, train, and deploy with reinforcement learning through managed reinforcement learning algorithms, support for multiple frameworks, simulation environments, and integration with AWS RoboMaker.
In just a few lines of code, developers can start learning about reinforcement learning with AWS DeepRacer, a 1/18th scale fully autonomous race car.
Amazon SageMaker Neo compiles models for specific hardware platforms, optimizing their performance automatically, allowing them to run at up to twice the performance, without any loss in accuracy. Thus, developers no longer need to spend time hand tuning their trained models for each and every hardware platform.
AI Services to all applications
Integrating machine learning into applications can be difficult for any developer. So, AWS wanted to help developers add machine learning to applications without any experience.
Amazon Textract uses machine learning to read almost any type of document to extract text and data without the need for any manual review or custom code. Also, it allows developers to automate document workflows, processing millions of document pages in a few hours.
Comprehend Medical is a highly accurate natural language processing service for medical text. It uses machine learning to extract disease conditions, medications, and treatment outcomes from patient notes, clinical trial reports, as well as other electronic health records.
Amazon Personalize is a real-time recommendation and personalization service. So, users can create recommendation models with the power of a fully managed Amazon service.
Like Amazon Personalize, Amazon Forecast is based on technology that has been developed by Amazon.com and used for a lot of critical forecasting. It creates accurate time-series forecasts using historical and related causal data. The subsequent machine learning models help users create an optimal customer experience.
Managed Service Providers
Managed service providers will help enterprises take advantage of these new capabilities. Although many were designed for ease of use, MSPs offer a full-on machine learning assistance and capabilities outside of Amazon’s specific offerings.