Amazon Web Services is offering machine learning algorithms and model packages on their AWS Marketplace. This was announced at AWS re:Invent Conference last week. You can choose from free and paid algorithms and models from various categories such as computer vision (CV), natural language processing (NLP), speech recognition, text, data, voice, image, video analysis, and predictive analysis.
Shaun Ray wrote in the AWS News Blog about the details of this offering, explaining how developers can use the ML algorithms and deploy them directly on Amazon SageMaker service. Amazon SageMaker, a fully-managed machine learing service launched last year, provides developers and data scientists the ability to build, train, and deploy machine learning models.
SageMaker uses MXNet, TensorFlow, PyTorch, and Chainer machine learning and deep learning frameworks and covers the machine learning workflow: label and prepare your data, choose an algorithm, train the algorithm, tune and optimize it for deployment, make predictions, and take action.
The AWS Marketplace Machine Learning offering includes 150+ algorithms and model packages with a selection for different industries like retail, media, manufacturing, and HCLS. Customers can find solutions to a variety of use cases like breast cancer prediction, loan risk prediction, vehicle recognition, botnet attack detection, automotive telematics, motion detection, demand forecasting, and speech recognition.
Developers can browse the list of algorithms, select and subscribe to a machine learning solution, and deploy it using one of the following tools:
- SageMaker console
- Jupyter Notebook
- SageMaker SDK
- AWS Command Line Interface (AWS CLI)
Ray discussed how to use AWS machine learning algorithms by trying the Deep Vision vehicle recognition model from Deep Vision AI. This model allows developers to identify the car details like vehicle make, model and type based on a set of uploaded images. The result output includes an attribute called score, a measure of how confident the model is about the result. The range of the score is 0.0 to 1.0.
At launch, AWS Marketplace for Machine Learning includes algorithms and models from companies like Intel, H2O.ai, Deep Vision AI, Cloudwick Technologies, and Knowledgent, and Persistent Systems.
In terms of security, AWS Marketplace provides static scans, network isolations, validation, and catalog curation on algorithms and models for Amazon SageMaker to help keep the data secure. The data is also protected by encrypting the algorithms and model package artifacts in transit and at rest, using secure (SSL) connections for communications, and via role based access control.
Customers pay a subscription fee for the use of an algorithm or model package and the AWS resource fee. AWS Marketplace provides a consolidated monthly bill for all purchased subscriptions. More information on how to deploy the machine learning models to Amazon SageMaker can be found in this guide.
In a related news, Amazon also announced last week that the machine learning courses used to train engineers at Amazon are now available to all developers through AWS.