Things are heating up in the cloud computing space as service providers now have a new arena for battle: artificial intelligence.
Amazon is the biggest online retailer in the world, but it is also one of the leading cloud computing companies through Amazon Web Services (AWS), an enviable position it shares with Google and Microsoft. Amazon recently joined the artificial intelligence fight.
Amazon Web Services is fastest growing computing and data storage service, making it the most profitable Amazon business. According to CNBC, AWS earned $3.23 billion for the third quarter of 2016, an increase of 55% compared to the previous quarter.
Now, Amazon is looking to beef up its cloud services with artificial intelligence offerings that include three new services and by investing in MXNet. MXNet is a deep learning framework that will soon be behind its artificial intelligence offerings in the future.
AI as a cloud computing technology is already in use at Amazon and you could see it in several aspects of their business. These AI systems help identify movie scenes, getting orders fulfilled and even in the voice recognition service, Alexa. The retail giant is taking it a step further, catching up with Microsoft, Facebook, and Google who have led the research into AI for many years now.
The company has revealed that they have thousands of engineers who are hard at work and focused on artificial intelligence. The company has yet to reveal their AI programs, but it has published a handful of white papers and scientific papers that touched on its foray into artificial intelligence.
Amazon does have three artificial intelligence offerings available now.
1. Amazon Rekognition. An image detection service that uses deep learning. Deep learning utilizes a mix of feature abstraction and computing power to distinguish images. Amazon Rekognition is a fully-managed offering that can determine objects, faces and scenes. If you send an image to Amazon Rekognition, it will come up with a variety of labels and corresponding confidence level that it is correct. For instance, if you send a photo of a cat, it will return tags such as cat, animal, Persian cat, and pet. When detecting faces, Rekognition will try to describe it for you, such as whether the person is smiling, appears to be sad, wearing eyeglasses, closed mouth and other features. This is a great service for those who have lots of photos that they want to index. The service can process even millions photos a day.
Another use case for Rekognition is in authentication and security. For instance, you can use this to give or deny access to your employees. They look into a camera, which would then send that image to your application. Rekognition will compare that image with your employee’s photo on file and unlock the door if it matches. Or you can do visual surveillance.
2. Amazon Lex. Amazon also unveiled Lex, which uses automatic speech recognition and natural language understanding that you find on Amazon Alexa. Lex can be used to build web and mobile applications such as chatbots, which mimic humans when interacting with you. These human-like chatbots are currently compatible with Facebook Messenger. Several other assimilations are underway, including integration for Twilio and Slack.
3. Amazon Polly. Amazon Polly is the third artificial intelligence offering that Amazon launched in November 2016. Polly reads text in a lifelike way. You can use Amazon Polly for your own applications and tools. Right now, Polly has close to 4-dozen voices and can read text in 2-dozen languages. More voices and languages are in the works. Polly can distinguish between two words. For example, it will pronounce “minute” correctly depending on the context in which the word appears, such as “It would take a minute to process your order.” versus “Just add a minute amount of salt.”
Polly will also say “Live from Seattle” differently from “I live near the mall.” Aside from correctly pronouncing words according to the context they are used in, Polly can also correctly read out abbreviations (such as St. for street vs. saint), fractions, currencies, times, dates, abbreviation and just about any element of speech.
Polly can process five million characters a month without paying for anything. After the free allocation, however, you would need to pay a rate of $0.000004 per character. That’s four cents for every 10,000 characters. Polly is very secure as it encrypts all the data it gets while transferring audio using SSL.
The launch of these artificial intelligence cloud services for customers will be a first for Amazon.
It is not the first time that cloud computing technology is used in Amazon’s other products. In fact, one of the biggest challenges for the company is to bring together the small teams that worked on these technologies and for different services together.
The good news is that these AI cloud computing services are their own – you don’t have to get these from third party companies.
Expect more AI cloud solutions with MXNet
Amazon has also announced investing in MXNet, which is a scalable deep learning platform that can support cutting edge deep models, including long short-term memory networks and convolutional neural networks. Developers can now take advantage of CloudFormation and Deep Learning AMIs that AWS offers. They can also benefit from broad support and an existing library of sample tutorials, as well as support from a thriving community.
Catching up with other cloud computing companies
It is a great move for Amazon to offer artificial intelligence for their cloud computing services. More and more companies are going to demand AI for their own applications. What’s more, AI works with complex algorithms, which means that it will need more compute resources.
In short, artificial intelligence for their cloud services would mean more business for them.
Four Cornerstone can help you get on Amazon Web Services. Call us at +1 (817) 377-1144 and find out how we could be a partner if you should ever decide on migrating your applications to the cloud.
Photo courtesy of A Health Blog.