Why ML is Perfect for Lazy Developers

  • by
Why ML is Perfect for Lazy Developers

Why ML is perfect for lazy developers – The Microsoft AI avocado, Henk Boelman explains why artificial intelligence is perfect for every lazy developer. Henk explains how Azure Cognitive Services can detect faces and interpret the feelings on that face, identify objects in pictures and how you can use it with any project you’re working on. Mmm avocados.

Today we’re going to talk about obviously machine learning but also more of a sort of where does that fit into the Azure kind of whole picture of everything, right? So Henk, why machine learning? It’s like a great way of doing things and actually if you’re like a lazy developer, you don’t have to type all those “if and else” statements. You just make your data scientists give you data and you can actually train a model, but even if you don’t want to do that, don’t want to train it yourself.

There are a lot of cognitive services which you can just use, just call an API and you have intelligence in your application. So our first layer is cognitive services layer. There are like around 30 APIs, you can just leverage your solutions. So there’s an API. If you want to do face detection, you send a picture, get all the faces back, tell you about your age, emotions, your gender, if you’re happy or not.

Or if you want to do like main object detection, like this is a car, this is a person you send a photo.. it will even tell you where those objects are and that’s just an API call. So any developer should be able to do that in any language. These machine learning models are very general for the object detection, it’s very general.

It can detect a car but it doesn’t, cannot detect a brand or very specific things. So then when you, when you want to make them more specific, we have surfaces as the custom vision service where you can upload your own, your pictures just through an interface upload like 50 or more from your object, hit a train button and as well expose a model for you. So then you can basically detect everything you want. So you said model I noticed. Can you, what is a model in machine learning world?

Like if I’m a developer, I’ve never looked at machine learning. How do you explain a model? What is it? So a model is the delivery of your training. So you train a model using TensorFlow or our custom vision service. Basically it is like a method. Something goes in, something goes out only. Only now you didn’t explicitly program it, but the algorithm created this model for you. So from a developer’s perspective, you can look at it as a function.

You put X in, you get Y out. Okay. If you want to do it yourself on your own data on scale, we have Azure machine learning service and that tool enables you to work as a team together to build predictive and analytics solutions, work as a team. You start building your pipelines. Like Damian always talks about about ML Ops Yeah, that’s Damian Brady who was also on the show.

Yes. So he talks about that and that process delivers a model, but to run those pipelines, to train your model that needs so much power. So models train for like two weeks or like these big GPU clusters we have. And your small laptop will just melt. Yeah, I have a project on how you can create a smile detector and that will give you like a good introduction of what is possible with the face API.

Oh, that’s very cool. So the link will be below, so go check that out as well. Yes. But what about, is that docs on Azure as well? Yes, there are lots of MS Learn content. There are a lot of docs with lots of great tutorials and of course we travel all over the world to talk about how to get started. Which is why you’re here. Thank you.



Useful links:

reference – Why ML is perfect for lazy developers

Share this post ...

Leave a Reply

Your email address will not be published. Required fields are marked *