Amazon built its retail success in part on the successful application of machine learning to the consumer experience. Now it is taking all that corporate learning and making it available to its AWS clients via a plethora of new services it announced at it’s re:invent conference in Las Vegas this morning.
If it is successful it could well democratise access to machine learning in much the same way it has made enterprise-grade computing ubiquitous.
In later comments to media, AWS CEO Andrew Jassy described the announcement of Amazon Sagemaker as a gamechanger as it would make it possible for a much wider set of developers to machine learning which today is held back by a lack of capabilities in the market. Salaries north of $200,000 are not uncommon in markets like Australia, because of lack of skilled practitioners.
“You also need a broad set of components in machine learning to be successful,” he said. “But those services don’t really matter unless you have the underlying infrastructure to offer those services.”
Actually, the machine learning news represents just a few out of more than 20 announcements at the event. More of these later.
According to Jassy his company wants to sweep away many of the difficulties developers – or builders in AWS parlance – encounter.
“They don’t want it to be so difficult, they don’t want it to be a black box, and they do want it to be much easier to work with,” he said.
“Machine learning is so tantalising for developers and data scientists, but there are a lot of constraints.”
He also outlined the many ways Amazon is already using machine learning for its own purposes.
“Most of the things you see in our consumer business are fuelled by machine learning such as recommendation engines.”
To that, you can also add the way it has optimised the robots in its fulfillment centers, the natural language and speech recognition in Alexa and even the work it has done on drones, he said.
AWS it taking the internal expertise garnered from years of real-world deployment to provide tools and services to help its clients to accelerate their own machine learning programs.
There is huge interest in machine learning among the customer set, Jassy said and he claimed five times as many AWS customers are working on machine learning compared to it rivals (although exactly how he arrived at that number is a bit opaque).
“Yet,” he said, “it is very early for most customers especially in mainstream enterprises.”
To help accelerate the adoption of machine learning Jassy announced Amazon Sagemaker.
According to the company, “Amazon SageMaker is a fully managed end-to-end machine learning service that enables data scientists, developers, and machine learning experts to quickly build, train, and host machine learning models at scale. This drastically accelerates all of your machine learning efforts and allows you to add machine learning to your production applications quickly.”
There’re three core components to Sagemaker;
• Authoring: Zero-setup hosted Jupyter notebook IDEs for data exploration, cleaning, and preprocessing. These can be run on general instance types or GPU powered instances.
• Model Training: A distributed model building, training, and validation service. You can use built-in common supervised and unsupervised learning algorithms and frameworks or create your own training with Docker containers. The training can scale to tens of instances to support faster model building. Training data is read from S3 and model artifacts are put into S3. The model artifacts are the data dependent model parameters, not the code that allows you to make inferences from your model. This separation of concerns makes it easy to deploy Amazon SageMaker trained models to other platforms like IoT devices.
• Model Hosting: A model hosting service with HTTPs endpoints for invoking models to get real-time inferences. These endpoints can scale to support traffic and allow you to A/B test multiple models, simultaneously. Again, you can construct these endpoints using the built-in SDK or provide your own configurations with Docker images.
AWS also announced a series of machine learning services which companies can purchase.
For starters it will now extend it’s Amazon rekogniton service which it announced last year to video while at the same time announcing a new service called Amazon Kinesis for video stream ingestion.
It has also released a transcription service, a translation service and Amazon Comprehend, a fully managed natural language service.
According to a company blog post from this morning, “Amazon Comprehend analyses text and tells you what it finds… It can identify different types of entities (people, places, brands, products, and so forth), key phrases, sentiment (positive, negative, mixed, or neutral), and extract key phrases, all from the text in English or Spanish. Finally, Comprehend‘s topic modeling service extracts topics from large sets of documents for analysis or topic-based grouping.”
The service has four key functions to start with; language detection, entity categorisation, sentiment analysis, and key phrase extraction.