Amazon SageMaker
{{Advert|date=September 2023}}
{{short description|Cloud machine-learning platform}}
{{Infobox software
| name = Amazon SageMaker AI
| developer = Amazon, Amazon Web Services
| released = {{Start date and age|2017|11|29|df=yes}}
| genre = Software as a service
| website = {{URL|https://aws.amazon.com/sagemaker}}
}}
Amazon SageMaker AI is a cloud-based machine-learning platform that allows the creation, training, and deployment by developers of machine-learning (ML) models on the cloud. It can be used to deploy ML models on embedded systems and edge-devices.{{Cite news|url=https://towardsdatascience.com/with-these-new-additions-aws-sagemaker-is-starting-to-look-more-real-b60f95bcbc38|title=With These New Additions, AWS SageMaker is Starting to Look More Real for Data Scientists|last=Rodriguez|first=Jesus|date=2018-11-30|work=Towards Data Science|access-date=2019-06-09}}{{Dead link|date=August 2023 |bot=InternetArchiveBot |fix-attempted=yes }}{{Cite news|url=https://www.fastcompany.com/90246028/how-ai-is-helping-amazon-become-a-trillion-dollar-company|title=How AI is helping Amazon become a trillion-dollar company|last=Terdiman|first=Daniel|date=2018-10-05|work=Fast Company|access-date=2019-06-09}} The platform was launched in November 2017.
Capabilities
SageMaker enables developers to operate at a number of different levels of abstraction when training and deploying machine learning models. At its highest level of abstraction, SageMaker provides pre-trained ML models that can be deployed as-is.{{Cite news|url=https://aws.amazon.com/blogs/machine-learning/deploy-trained-keras-or-tensorflow-models-using-amazon-sagemaker/|title=Deploy trained Keras or TensorFlow models using Amazon SageMaker|last=Ponnapalli|first=Priya|date=2019-01-30|work=AWS|access-date=2019-06-09}} In addition, it offers a number of built-in ML algorithms that developers can train on their own data.
The platform also features managed instances of TensorFlow and Apache MXNet, where developers can create their own ML algorithms from scratch. Regardless of which level of abstraction is used, a developer can connect their SageMaker-enabled ML models to other AWS services, such as the Amazon DynamoDB database for structured data storage,{{Cite news|url=https://searchaws.techtarget.com/tip/Evaluate-when-to-use-added-AWS-Step-Functions-actions|title=Evaluate when to use added AWS Step Functions actions|last=Marquez|first=Ernesto|work=TechTarget|access-date=2019-06-09}} AWS Batch for offline batch processing,{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/11/aws-step-functions-adds-eight-more-service-integrations|title=AWS Step Functions Adds Eight More Service Integrations|date=2018-11-29|work=AWS|access-date=2019-06-09}} or Amazon Kinesis for real-time processing.{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/08/deploy-sagemaker-and-a-data-lake-on-aws-with-new-quick-start|title=Deploy Amazon SageMaker and a Data Lake on AWS for Predictive Data Science with New Quick Start|date=2018-08-15|work=AWS|access-date=2019-06-09}}
Development interfaces
A number of interfaces are available for developers to interact with SageMaker. First, there is a web API that remotely controls a SageMaker server instance.{{Cite news|url=https://aws.amazon.com/blogs/machine-learning/call-an-amazon-sagemaker-model-endpoint-using-amazon-api-gateway-and-aws-lambda/|title=Call an Amazon SageMaker model endpoint using Amazon API Gateway and AWS Lambda|last=Olsen|first=Rumi|date=2018-07-19|work=AWS|access-date=2019-06-09}} While the web API is agnostic to the programming language used by the developer, Amazon provides SageMaker API bindings for a number of languages, including Python, JavaScript, Ruby, Java, and Go.{{Cite web|url=https://aws.amazon.com/sagemaker/developer-resources|title=Amazon SageMaker developer resources|website=AWS|access-date=2019-06-09}}{{Cite news|url=https://venturebeat.com/2018/11/21/amazon-updates-sagemaker-with-new-built-in-algorithms-and-git-integration|title=Amazon updates SageMaker with new built-in algorithms and Git integration|last=Wiggers|first=Kyle|date=2018-11-21|access-date=2019-06-09}} In addition, SageMaker provides managed Jupyter Notebook instances for interactively programming SageMaker and other applications.{{Cite web|url=https://docs.aws.amazon.com/sagemaker/latest/dg/nbi.html|title=Use Notebook Instances|website=AWS|access-date=2019-06-09}}{{Cite news|url=https://www.forbes.com/sites/forbestechcouncil/2018/08/17/here-come-the-notebooks/#6ec3e3b77609|title=Here Come The Notebooks|last=Gift|first=Noah|date=2018-08-17|work=Forbes|access-date=2019-06-09}}
History and features
- 2017-11-29: SageMaker is launched at the AWS re:Invent conference.{{Cite news|url=https://techcrunch.com/2017/11/29/aws-releases-sagemaker-to-make-it-easier-to-build-and-deploy-machine-learning-models/|title=AWS releases SageMaker to make it easier to build and deploy machine learning models|last=Miller|first=Ron|date=2017-11-29|work=TechCrunch|access-date=2019-06-09}}{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2017/11/introducing-amazon-sagemaker/|title=Introducing Amazon SageMaker|date=2017-11-29|work=AWS|access-date=2019-06-09}}{{Cite news|url=https://www.datanami.com/2017/11/29/aws-takes-muck-ml-sagemaker|title=AWS Takes the 'Muck' Out of ML with SageMaker|last=Woodie|first=Alex|date=2017-11-29|work=datanami|access-date=2019-06-09}}
- 2018-02-27: Managed TensorFlow and MXNet deep neural network training and inference are now supported within SageMaker.{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/02/amazon-sagemaker-now-supports-tensorflow-1-5--apache-mxnet-1-0--and-cuda-9-for-p3-instance-optimization/|title=Amazon SageMaker now supports TensorFlow 1.5, Apache MXNet 1.0, and CUDA 9 for P3 Instance Optimization|date=2018-02-27|work=AWS|access-date=2019-06-09}}{{Cite news|url=https://www.oreilly.com/ideas/how-to-jump-start-your-deep-learning-skills-using-apache-mxnet|title=How to jump start your deep learning skills using Apache MXNet|last=Roumeliotis|first=Rachel|date=2018-03-07|work=O'Reilly|access-date=2019-06-09}}
- 2018-02-28: SageMaker automatically scales model inference to multiple server instances.{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/02/auto-scaling-in-amazon-sagemaker-is-now-available/|title=Auto Scaling in Amazon SageMaker is now Available|date=2018-02-28|work=AWS|access-date=2019-06-09}}{{Cite news|url=https://polarseven.com/amazon-sagemaker-now-uses-auto-scaling|title=Amazon Sagemaker Now Uses Auto-scaling|date=2018-03-24|work=Polar Seven|access-date=2019-06-09}}
- 2018-07-13: Support is added for recurrent neural network training, word2vec training, multi-class linear learner training, and distributed deep neural network training in Chainer with Layer-wise Adaptive Rate Scaling (LARS).{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/07/amazon-sagemaker-announces-enhancements-for-built-in-algorithms-and-frameworks/|title=Amazon SageMaker Announces Several Enhancements to Built-in Algorithms and Frameworks|date=2018-07-13|work=AWS|access-date=2019-06-09}}{{Cite news|url=https://pureai.com/articles/2018/07/16/aws-sagemaker-update.aspx|title=Amazon Updates SageMaker ML Platform Algorithms, Frameworks|last=Nagel|first=Becky|date=2018-07-16|work=Pure AI|access-date=2019-06-09}}
- 2018-07-17: AWS Batch Transform enables high-throughput non-real-time machine learning inference in SageMaker.{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/07/amazon-sagemaker-supports-high-throughput-batch-transform-jobs-for-non-real-time-inferencing|title=Amazon SageMaker Now Supports High Throughput Batch Transform Jobs for Non-Real Time Inferencing|date=2018-07-17|work=AWS|access-date=2019-06-09}}{{Cite news|url=https://medium.com/@julsimon/making-the-most-of-your-machine-learning-budget-on-amazon-sagemaker-a6982bdd5edd|title=Making the most of your Machine Learning budget on Amazon SageMaker|last=Simon|first=Julien|date=2019-01-24|work=Medium|access-date=2019-06-09}}
- 2018-11-08: Support for training and inference of Object2Vec word embeddings.{{Cite news|url=https://aws.amazon.com/blogs/machine-learning/introduction-to-amazon-sagemaker-object2vec/|title=Introduction to Amazon SageMaker Object2Vec|date=2018-11-08|work=AWS|access-date=2019-06-09}}{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/11/amazon-sagemaker-now-supports-object2vec-and-ip-insights-built-i|title=Amazon SageMaker Now Supports Object2Vec and IP Insights Built-in Algorithms|date=2018-11-19|work=AWS|access-date=2019-06-09}}
- 2018-11-27: SageMaker Ground Truth "makes it much easier for developers to label their data using human annotators through Mechanical Turk, third-party vendors, or their own employees."{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/11/introducing-amazon-sagemaker-groundtruth|title=Introducing Amazon SageMaker Ground Truth - Build Highly Accurate Training Datasets Using Machine Learning|date=2018-11-28|work=AWS|access-date=2019-06-09}}
- 2018-11-28: SageMaker Reinforcement Learning (RL) "enables developers and data scientists to quickly and easily develop reinforcement learning models at scale."{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/11/amazon-sagemaker-announces-support-for-reinforcement-learning|title=Introducing Reinforcement Learning Support with Amazon SageMaker RL|date=2018-11-28|work=AWS|access-date=2019-06-09}}
- 2018-11-28: SageMaker Neo enables deep neural network models to be deployed from SageMaker to edge-devices such as smartphones and smart cameras.{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2018/11/introducing-amazon-sagemaker-neo|title=Introducing Amazon SageMaker Neo - Train Once, Run Anywhere with up to 2x in Performance Improvement|date=2018-11-28|work=AWS|access-date=2019-06-09}}
- 2018-11-29: The AWS Marketplace for SageMaker is launched. The AWS Marketplace enables 3rd-party developers to buy and sell machine learning models that can be trained and deployed in SageMaker.{{Cite news|url=https://www.fiercetelecom.com/telecom/aws-goes-deep-and-wide-machine-learning-services-and-capabilities|title=AWS goes deep and wide with machine learning services and capabilities|last=Robuck|first=Mike|date=2018-11-29|work=FierceTelecom|access-date=2019-06-09}}
- 2019-01-27: SageMaker Neo is released as open-source software.{{Cite news|url=https://www.forbes.com/sites/janakirammsv/2019/01/27/amazon-open-sources-sagemaker-neo-to-run-machine-learning-models-at-the-edge/#260498d14d03|title=Amazon Open Sources SageMaker Neo To Run Machine Learning Models At The Edge|last=Janakiram|first=MSV|date=2019-01-27|work=Forbes|access-date=2019-06-09}}
Notable Customers
- NASCAR is using SageMaker to train deep neural networks on 70 years of video data.{{Cite news|url=https://www.zdnet.com/article/nascar-to-migrate-18-petabytes-of-video-archives-to-aws/|title=NASCAR to migrate 18 petabytes of video archives to AWS|last=Digman|first=Larry|date=2019-06-04|work=ZDNet|access-date=2019-06-09}}
- Carsales.com uses SageMaker to train and deploy machine learning models to analyze and approve automotive classified ad listings.{{Cite news|url=https://www.itnews.com.au/news/carsales-builds-tessa-ai-to-check-vehicle-ads-524583|title=Carsales builds Tessa AI to check vehicle ads|last=Crozier|first=Ry|date=2019-05-02|work=IT News|access-date=2019-06-09}}
- Avis Budget Group and Slalom Consulting are using SageMaker to develop "a practical on-site solution that could address the over and under utilization of cars in real-time using an optimization engine built in Amazon SageMaker."{{Cite news|url=https://aws.amazon.com/about-aws/whats-new/2019/05/avis-slalom-machine-learning-on-aws|title=Avis Budget Group and Slalom Further Digitize the Car Rental Process with Machine Learning on AWS|date=2019-05-31|work=AWS|access-date=2019-06-09}}
- Volkswagen Group uses SageMaker to develop and deploy machine learning in its manufacturing plants.{{Cite news|url=https://metrology.news/volkswagen-and-aws-join-forces-to-transform-automotive-manufacturing/|title=Volkswagen and AWS Join Forces to Transform Automotive Manufacturing|date=2019-05-24|work=Metrology News|access-date=2019-06-09|archive-date=2020-10-28|archive-url=https://web.archive.org/web/20201028173020/https://metrology.news/volkswagen-and-aws-join-forces-to-transform-automotive-manufacturing/|url-status=dead}}
- Peak and Footasylum use SageMaker in a recommendation engine for footwear.{{Cite news|url=https://www.computerweekly.com/news/252463301/Footasylum-steps-up-artificial-intelligence-to-drive-customer-centricity|title=Footasylum steps up artificial intelligence to drive customer centricity|last=Mari|first=Angelica|date=2019-05-14|work=Computer Weekly|access-date=2019-06-09}}
Awards
In 2019, CIOL named SageMaker one of the "5 Best Machine Learning Platforms For Developers," alongside IBM Watson, Microsoft Azure Machine Learning, Apache PredictionIO, and AiONE.{{Cite news|url=https://www.ciol.com/5-best-machine-learning-platforms-developers|title=5 Best Machine Learning Platforms For Developers|last=Pandey|first=Ashok|date=2019-02-21|work=CIOL|access-date=2019-06-09}}