Since 2006, Amazon Web Services (AWS) has set itself the goal of helping developers and data scientists build their own Machine Learning and Artificial Intelligence services. And to do so without their needing to be experts in the field. On both fronts, AWS is moving very fast. Within its “Development, Serverless, AI and Machine Learning” division, Premaccess supports its clients who are ready to get to grips with these AWS tools. Falco Schmutz, CEO, explains how.
Amazon Web Service is today a real driving force in Machine Learning and Artificial Intelligence. Its progress in these sectors is beyond dispute. Google was a pioneer in these fields, but AWS has gradually caught up with it. Both are educating the market in the use of these large-scale technologies.
Having established itself in the Cloud Computing market, Amazon Web Service intends to do the same with machine learning and AI. At its most recent major events, in 2018, the company stepped up its communication along these lines.
TABLE OF CONTENTS
- What is machine learning?
- Artificial intelligence, effective and accessible with AWS
- How does Premaccess support its clients in machine learning and AI?
- The aws marketplace dedicated to artificial intelligence
- Launch your applications on AWS with Premaccess
What is machine learning?
Machine Learning is based on a series of algorithms capable of producing mathematical models that can be used when developing an application or a service. Today this technology can be used, in particular, to:
- detect fraud: the predictive models put in place with Machine Learning can help identify potentially fraudulent transactions.
- produce recommendations: depending on how a visitor browses your site, Machine Learning will interpret their searches in order to offer them the most personalised user experience possible and a series of recommendations close to their interests.
- create targeted marketing campaigns: Machine Learning can also be used to create targeted mailing campaigns. Depending on a visitor’s activity on your site, you can choose to send them relevant mailing campaigns based on their most recent searches.
- model behaviour: Machine Learning makes it possible to model the behaviour of a complex industrial process in order to identify deviations that will require maintenance action.
Artificial intelligence, effective and accessible with AWS
AI, for its part, will identify trends and produce predictions on the basis of this series of algorithms.
To that end, AWS has several application services. All are accessible over API and provide ready-to-use “intelligence”. We can mention in particular:
- Amazon Rekognition. With this service, you can add image or video analysis to your applications. It can identify objects, people, text, scenes and even activities. It is able to provide accurate facial recognition on your image and video files, and to detect any inappropriate content.
- With Amazon Polly, a text to speech application (25 languages, 53 voices, in real time), you can convert text into realistic audio recordings using deep learning. This service lets you create applications that “speak” thanks to a speech synthesis service.
- Translate is a real-time translation service. Accessible via API, it automatically recognises the source language.
- Transcribe is a speech to text service. It can generate a text file from an audio file. This application recognises text and punctuation, and detects the different speakers. It supports files of good or poor audio quality, including telephone quality. Its main use case today is call centres.
- The Comprehend service processes natural language to extract entities, people, places, product names, organisations, key phrases and so on from several documents (email, comments, social network posts). With Comprehend, machine learning is particularly effective at identifying specific points of interest in large sets of text (identifying company names in analyst reports, for example). It also makes it possible to grasp the sentiment behind the language used (identifying negative reviews…). The Washington Post uses it, for example, to create topic groups on its very specific articles.
How does Premaccess support its clients in machine learning and AI?
Today AWS is seeking to make Machine Learning and Artificial Intelligence mainstream among businesses. It is succeeding with ease.
The strength of AWS is being able to provide micro-service building blocks that are simple, solid and easy to grasp. “Thanks to these building blocks, we can easily support our clients in developing what they need and enable them to get to grips with these technologies, explains Falco Schmutz, CEO of Premaccess. We can, for example, set up small labs to show them how it works, at low cost.”
Nevertheless, to do something interesting with these AWS services, you need a concept, an idea.
When Premaccess supports a client, the first step is to properly define its project, so that it is aligned with its business.
The second step, just as important, is to define which data will be used to build the product the clients have asked for (image, video, audio, data, etc.). Of all the available data, which can be labelled correctly, which can be easily interpreted by the AWS Machine Learning and Artificial Intelligence tools?
Faced with these two technologies, Premaccess always tries to bring these clients back to a business use case. The key question is: “What do you need? What causes problems at your end, what is time-consuming, that we could try to solve with Machine Learning and/or Artificial Intelligence tools?”
Premaccess opened, six months ago, a new “Development, Serverless, AI and Machine Learning” division made up today of four people with expertise in AWS who regularly meet client needs. “We currently have requests for comment moderation, image analysis, and for setting up predictive models and decision trees”, reports Falco Schmutz.
Premaccess
The AWS marketplace dedicated to artificial intelligence
Last November, AWS kept up the momentum by presenting its marketplace dedicated to AI. The event took place in Las Vegas.
On this platform, close to 150 Machine Learning algorithms and models are offered by AWS itself and some forty partner software vendors. Among the latter we find service providers (LexisNexis) and technology suppliers (Intel).
Accessible from Amazon SageMaker, these algorithms and models are split into different categories: Computer vision, Natural language processing, Speech recognition, Image, Text, Structured, Audio, Video. All have the advantage of being usable in “Pay as you go” mode, that is, billed on usage. And all make it possible, once again, to build applications from machine learning software building blocks.
This marketplace confirms AWS’s ambition to consolidate its place in Cloud AI. It is evolving fast at the moment, and clearly sets the trends. It is worth following closely.
Launch your applications on AWS with Premaccess
If you want to use AWS technologies to design and manage your applications, the Premaccess team is at your disposal.
It will answer all your questions, study your needs, and support you step by step in setting up your web services.
In addition, once deployed, your applications will be monitored by our experts to ensure their evolution and their level of security over the long term.