Quick note: I’ve started a weekly email newsletter covering all things deep learning and self-driving cars. I call it Transmission, sign up today!

Deep Learning is Revolutionary

10 reasons why deep learning is living up to the hype

Many have written about how deep learning is taking over the world and why that is important; I cannot echo them enough. Playing with deep learning is the closest I’ve ever felt to being a magician, and it’s become clear to me that every (great) piece of software will be powered by deep learning within the next ~3 years. However, deep learning isn’t mainstream yet, so I thought I’d share work by some very talented contributors, in the hopes to bring it just that little bit closer.

Here’s ten reasons why I think deep learning is living up to the hype…

1 Stuck with a low-resolution photo? Deep learning can predict what the higher-resolution photo might look like, and add missing details

…it’s equally impressive for anime art

2 Deep learning has enabled a text-to-speech system that is almost indistinguishable from human voice. Think of the possibilities!

3 Deep learning can compose classical music that you’d believe to be created by a human

4 Deep learning can replicate the style of your favorite painter with the image of your choice

5 Deep learning can even generate fonts!

6Deep learning can auto-fill missing parts of an image, by predicting what should be in that space

7Deep learning can train a robot to walk like a human

…or train a robot to grasp objects like a human

8Deep learning can caption an image, just like a human would

9Want to sketch a beautiful landscape but can’t draw? Don’t worry, deep learning can take it from here!

10Best of all (at least to me), you can train a deep neural network to steer a car, just like a human

…it even turns out Grand Theft Auto is a great simulation environment for training a self-driving car

Why am I so sure that deep learning is revolutionary? If you look closely at the above tasks that the machine is performing, you’ll see a common thread: creativity. A machine, albeit trained with human data, is able to demonstrate almost artistic traits.

We’ve spent decades building and refining machines to be computationally powerful. In the next decade I hope we discover what the human race will be empowered to do when our software thinks creatively.

Oliver Cameron (@olivercameron) | Twitter
The latest Tweets from Oliver Cameron (@olivercameron). I lead the self-driving car team at @udacity. Previously…twitter.com