A dummy’s guide to Deep Learning (part 1 of 3)
Yes, you can do it. Today.
Wait, what?
Deep learning? Isn’t it the mysterious technology behind the mighty AlphaGo? Isn’t it the brain of self-driving cars?
Well, yes it is. Deep learning is a branch of machine learning that has shown incredible results on very difficult tasks like recognizing objects from an image, understanding speech and languages, and of course, playing board games. A bunch of smartest people have been working on it for decades, and it’s absolutely state-of-the-art.
But I’m not a crazy scientist! How can I do something with it?
OK. Since you’ve clicked into a dummy’s guide, chances are that you are a curious dummy. So first of all, let me answer a few questions I know a curious dummy might ask.
What is deep learning exactly?
It’s software programs trying to mimic the human brain. The way it forms, the way it learns and the way it responds. It’s inspired by neuroscience. It can be used for plenty of AI tasks that are extremely difficult to program.
How much do I need to learn?
You do need a lot of deep domain knowledge to make any significant contribution to the academic field, but the truth is, thanks to many powerful open source libraries, building useful applications out of the existing research progress is much easier than most would expect.
Why would I want to learn about it?
Because it’s a gold mine right now. Take a look at this $1000 question on Quora: https://www.quora.com/What-technological-changes-will-create-the-most-opportunities-for-new-startups-over-the-next-2-3-years.
Whenever there’s a technology break through, there’s always a good opportunity to make use of the technology and build amazing applications that could never been done before. Deep learning has had a lot of significant research progress, but look around — how many applications can you see?
Do I need a huge data center?
Training complicated models to power something like Google image search do require large scale datasets and a lot of compute power. But for developing something that can run on mobile phones? Not so much.
If Google’s data center is like a human, our mobile phone is like an ant. You can’t grow a human brain and plant it into an ant anyway, right? But ants are still smart enough and they can do many things.
In part II, we’ll introduce some basic concepts of deep learning, and in part III we’ll show you how we can build a really cool deep learning application from scratch. No need to know much about machine learning at all — you’ll see how easy it is to get started, and you’ll be given all the resources you need to become an expert.
Thank you for reading! If you enjoyed the article, please recommend it by clicking on the little heart button below, or share it with your friends! Follow The Bleeding Edge for the coming part II & III, and stay up-to-date with latest technologies!
You might be also interested in my latest article about building AI bots for emails:
