After the first announcement of Amazon Personalize during AWS re:Invent last November, the service is now generally available for all AWS customers. With this service developers can add custom machine learning models to their application, including models for personalized product recommendations, search results and direct marketing, even if they don’t already have machine learning experience.
Amazon Personalize is a fully managed service in AWS that trains, tunes, and deploys custom machine learning models by provisioning the necessary infrastructure. The service will handle things like data processing, feature extraction, algorithm training and optimization, and hosting. As Nick Walsh, developer relations AI/ML at Amazon, said on Twitter:
Build recommendation systems to power your applications, while Personalize does all of the heavy lifting.
With Amazon Personalize, customers can leverage machine learning models generated and perfected by Amazon from over 20 years of data, ranging from personalized buying experiences and product discovery, to check out, behind an API. Through an API call, developers can stream data to be processed and examined by Amazon Personalize – which will then identify what is meaningful, select the right algorithms, and train and optimize a personalization model that is customized for the data. Note that the data developers stream is kept private and is entirely owned by the customer.
Source: https://aws.amazon.com/personalize/
Various AWS customers have already included Amazon Personalize in their application, such as Yamaha Corporation of America, Subway, Zola and Domino’s. Neville Hamilton, interim chief information officer at Subway, said in a Business Wire article:
Using Amazon Personalize, we can quickly deliver personalized recommendations for our endless varieties of ingredients and flavors to fit the unique lifestyles of our busy guests. Amazon Personalize lets our team use simple API calls to curate recommendations without requiring machine learning expertise.
With the GA release of Personalize, Amazon is further strengthening its ML-based services portfolio and is positioning the company in the booming Machine Learning as a Service market (MLaaS) - which is riding on growing proliferation of cloud-based technologies. Other public cloud vendors like Microsoft and Google are also offering MLaaS services to customers with Azure Machine Learning and Google Cloud AI.
Amazon Personalize will be available in the US East (Ohio), US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), Asia Pacific (Singapore), and EU (Ireland) region, with more to launch soon. Amazon will charge:
- five cents per GB of data uploaded to Amazon Personalize, and
- 24 cents per training hour used to train a custom model with their data
Lastly, any real-time recommendation requests are priced based on how many are uploaded, with discounts for larger orders. For more detailed information on pricing, see the pricing page.
