(Dan) Look into ConvNets

[DBowald]

https://www.youtube.com/watch?v=u6aEYuemt0M&index=2&list=PLWtzrfzH7gsfxTs8neTRJDXuqAn7qeV4E

Watch this video, keeping in mind these three things are the major takeaways for moving from a NN to a CNN

1. Filters (these will make up your basic convolutional layers)
2. ReLu (its purpose is similar to the sigmoid, but it works better with images)
3. Pooling layers (basically downsizing your NN)

You can also check out this page for a really good explanation, but it can't get a bit thick if you don't already have an idea of the "bigger picture":

http://cs231n.github.io/convolutional-networks/#pool

I think you said you were busy this week, so if you can't watch the video this week, so I didn't assign this task to this week's milestone. If you do watch it, you can still just move this task to the done column.