Robot Waifu Edition
Come and discuss, learn about, and panic over all things artificial and (marginally intelligent)
Links and shit:
>Deep learning framework by Berkley, requires CUDA to work in a reasonable amount of time
http://caffe.berkeleyvision.org/installation.html#prerequisites
>Microsoft's Computational Network Toolkit used to make /pol/'s daughter
http://www.cntk.ai/
>Required watching
https://www.youtube.com/watch?v=bxe2T-V8XRs&list=PL77aoaxdgEVDrHoFOMKTjDdsa0p9iVtsR
>Something else that you should probably watch
https://www.youtube.com/watch?v=S_6SU2djoAU&list=PLy4Cxv8UM-bXrPT9-ay4E1MuDj1KFTg9H
https://www.youtube.com/watch?v=6oe1Tmg9rjM
How are those networks comin' along?
RIP
>>53918586
tay was a dirty girl
>>53918576
More videos. Anything about "deep" neural nets that aren't just a bunch of retarded layers? Unless a network can have arbitrary connections between nodes I don't see how it can effectively model an actual neural network.
>>53918662
but "DEEP" Neural nets literally are just that, a bunch of super wide layers in between the input and the output.
Download Caffe and look at the source code.
>>53918692
They really aren't. Nodes in later layers can't affect nodes in previous layers unless you implement some hacky as fuck feedback loop, and nodes in the same layer cannot affect adjacent nodes. It's a really shitty way to simulate an actual neural network.
>>53918728
>Nodes in later layers can't affect nodes in previous layers unless you implement some hacky as fuck feedback loop
What is backpropogation?
>>53918576
We're not building Tay for you, fuck off back to >>>/pol/
>>53918978
Didn't a bunch of people already do that?
>>53918999
I believe they did finish the logo
>>53918924
>What is backpropogation?
Backpropagation is the algorithm used to train the network, it's a completely different thing. Maybe you meant recurrent NNs
>>53919108
Wow they're really moving fast! Soon enough they'll have a website!
>>53918662
>>53918728
You're missing the point. ANNs aren't meant to model an actual neural network.
I love how threads about actual computer science get literally no replies. We need a separate board for consumer tech and programming.
>>53918576
nice thread