DeepDreams is a research project aimed at creating an algorithm that can generate the most nonsensical, illogical, and yet somehow fascinating images using a neural network.
We've trained our network on a dataset of 10,000 images of abstract art, 5,000 cat pictures, and 3,000 random text files.
Our goal is to create a network that can produce output that's more interesting than a kindergartener's art project.
Current Status: Failed (but we're working on it)
1. Increase training dataset to include more cat pictures
2. Add more layers to the network
3. Convince our investors to give us more funding (good luck with that)