Getting Started with Deep Dream

Fri, Dec 29, 2017

Math

There’s alot to learn generally about DD, mainly: python, calculus, statistics, convolutional neural networks. start here

What does it look like?

Here’s a listing of different channels you can run DD through. Every layer has n number of channels. This lists all the layers: http://storage.googleapis.com/deepdream/visualz/tensorflow_inception/index.html And here’s the channels for mixed4a_5x5_bottleneck_pre_relu: http://storage.googleapis.com/deepdream/visualz/tensorflow_inception/mixed4a_5x5_bottleneck_pre_relu.html

I also manage two subreddits, http://reddit.com/r/deepdreamvideo and also http://reddit.com/r/deepdreamgirls The most popular subreddit is http://reddit.com/r/deepdream

Don’t know how to code? No problem

I run an OS project to make it easier to get into DD, called bind. I do it because I love DD, I think art and tech meet here, and the future of art includes ML, DD, DL, and I want do my part, get people excited about it.

I include on that github page everything you need to start making cool images or videos.

Popular RN

hands down, it’s style transer, neural style, fast neural style and the like. https://github.com/lengstrom/fast-style-transfer https://github.com/andersbll/neural_artistic_style

Check em out, there’s a huge amount of different approaches. Personally DD is my favorite, but most people have moved on (the algo was release in ‘15 by google engineer Alexander Mordvintsev https://en.wikipedia.org/wiki/DeepDream)