Your Apps Need Machine Learning, Add It in Minutes with Cognitive Services

The complexity of applications is ever growing, regardless whether they’re mobile, web, desktop, or other; more and more applications add requirements for image analytics, voice recognition, command identification, and so much more. Coding these functionalities from the ground up is interesting, challenging, exciting, but time consuming.

Besides, building this kind of functionality requires experts that need to master Python or R, and understand what models to use in a specific case – i.e. is this a case for supervised or unsupervised machine learning? When do I use classification vs regression algorithms? If I have some previous real-world data where one of the features with the most weight in the model has quite a few outliers, can I use it in a Bayesian Network, or should I use a Bayesian Hierarchical Model for transfer learning?

All these are fair questions, some slightly harder to answer than others, but questions you would have to face when building the machine learning models for your application. But, is that really necessary if you just need to know if an image has a dog in it? Or if some text has a few words that you want to moderate?

You shouldn’t need to invest a lot of time and effort into creating a machine learning model that will do something fairly common and simple, and you don’t. You can implement some powerful machine learning to solve some common problems in just minutes, by just deploying a cloud service that will handle everything for you. If you are facing some of the scenarios listed below, keep reading; I will tell you how Cognitive Services can help.

Cognitive Services can help if:

• You need to analyse the sentiment of the feedback coming from users to generate reports for management to read.
• You need automatic translation of the transcript of a meeting, and even some automatic audio translation for employees from other countries to understand.
• You need to understand voice and text commands that trigger actions for a virtual assistant.
• You need to recognise people based on their voice to authenticate them into your service.
• You need to extract text from images, even if the text is handwritten, to improve search in your notes app.
• You need to detect faces in a picture to be able to create a tailored video for someone that participated on the city’s marathon, with only the photos in which they appear in.
• You need to moderate text and images so your young users can safely use your service.
• You need to create a transcript of a video and its translation to many languages, and identify who appears and when in the video for your YouTube-competitor idea.
• You need to identify through a picture whether I’m buying coriander or parsley.
• You need a bot that automatically answers frequently asked questions through Messenger, Slack, and your own website.

So what are these magical Cognitive Services?

Cognitive Services is an array of machine learning services provided by Microsoft through Azure, its cloud computing platform. You literally only need to provision the service that you want, set some things up here and there, and make some REST requests to be able to implement machine learning functionality into your apps. If you know how to perform REST requests in your language of choice, whether that is Java, Swift, Python, C#, JavaScript, or whatever, you will be able to implement machine learning in minutes.

“If you know how to perform REST requests in your language of choice, you will be able to implement machine learning in minutes.”

How much do these services cost?

With these being cloud based services, they come at a price to have implemented into your apps. We cannot tell you exactly how much it will cost in your particular case as the cost will vary from service to service, and depending on the use you give it. For example, it is more expensive to store individual videos in the cloud and process them to find people in the video – and where in the video they appear – than simply returning a decimal value representing the sentiment for a small body of text. On the other hand, the monthly cost of storing just one video versus servicing billions of request for sentiment analysis would be much less.

However, the important thing that I want you to know right now is initially, all these services are entirely free. If you are just testing the functionality, or have very few users that make few requests a month, the free version of the services could easily work for you; just keep in mind that eventually, you will have to migrate to a pay-as-you-go subscription. This simply means that, depending on how the services are used, you will be charged accordingly.

Take the first steps

Now you know that Azure offers a wide array of services for machine learning to be used inside your apps, you will be able to find all the services listed below. The very first thing that you need to do is to go to the Azure website and create your account; you will even find that it offers you $200 (that’s US dollars) worth of credit for you to use with paid services for your first month.  You will have access to some services for free for an entire year, and lifetime access to a few services that are always free to use. Once you have your Azure subscription ready, it’s time to start using one of these Cognitive Services:

• Computer Vision
• Video Indexer
• Custom Vision
• Face API
• Content Moderator
• Speech to Text
• Text to Speech
• Separker Recognition
• Speech Translation
• Text Analytics
• Translator Text
• Bing Spell Check
• Language Understanding
• QnA Maker for Knowledge Bases
• A bunch of search services for visual, web, video, entities, images, and more

Improve the experience that you offer to your users, and separate yourself from the competition. Add Azure Machine Learning using Cognitive Services skills to your resume today!