AI learns about cats

#4
http://www.dailytech.com/Googles Unsupervised SelfLearning Neural Network Searches For Cat Pics/article25025.htm

What do you make of this?. While it's interesting I don't think it's really solved many problems in terms of understanding the brain
I found it quite hard to understand their paper, but it looks like an attempt to emulate the brains physiological process of weighting/calculating sensory inputs using silicone, to create an associative learning network from the ground up.

I'm not sure it's that important to be honest. It still only attempting to emulate how sensory data might be presented to the brain, and emulate how networks might work which associate that data.

Our brains physiological networks are spatial, and they present sensory data spatially as a simple associative network in space.

But in my opinion this is missing a 'temporal' side of the process. Those spatial networks I mentioned, create spatial patterns of sensory data when they are activated, and it is these spatial patterns of activation which interfere with themselves temporally (across time and space). So that a past brain state can interfere with a future brain state.
 
Top