Wednesday, June 27, 2012

Google's A.I. recognizes cats. Laugh while you can.

Google's brain module was trained on YouTube stills. From vast amounts of data, one image spontaneously emerged ...
Using large-scale brain simulations for machine learning and A.I. | Official Google Blog 
".. we developed a distributed computing infrastructure for training large-scale neural networks. Then, we took an artificial neural network and spread the computation across 16,000 of our CPU cores (in our data centers), and trained models with more than 1 billion connections.  
...  to our amusement, one of our artificial neurons learned to respond strongly to pictures of... cats ... it “discovered” what a cat looked like by itself from only unlabeled YouTube stills. That’s what we mean by self-taught learning... 
... Using this large-scale neural network, we also significantly improved the state of the art on a standard image classification test—in fact, we saw a 70 percent relative improvement in accuracy. We achieved that by taking advantage of the vast amounts of unlabeled data available on the web, and using it to augment a much more limited set of labeled data. This is something we’re really focused on—how to develop machine learning systems that scale well, so that we can take advantage of vast sets of unlabeled training data.... 
... working on scaling our systems to train even larger models. To give you a sense of what we mean by “larger”—while there’s no accepted way to compare artificial neural networks to biological brains, as a very rough comparison an adult human brain has around 100 trillion connections.... 
..  working with other groups within Google on applying this artificial neural network approach to other areas such as speech recognition and natural language modeling."
Hah, hah, a cat. That's so funny. Unless you're a mouse of course.

The mouse cortex has 14 million neurons and a maximum of 45K connections per neuron, so ballpark estimate, perhaps 300 billion connections (real estimates are probably known from the mouse connectome project but I couldn't find them). So in this first pass Google has less than 1% of a mouse connectome.

Assuming they double the connectome every two years, they should hit mouse scale in nine years, or around 2021. There's a good chance you and will still be around then.

I've long felt that once we had a "mouse-equivalent" connectome we could probably stop worrying about global warming, social security, meteor impacts, cheap bioweapons, and the Yellowstone super volcano.

Really, we're just mice writ large. That cat is looking hungry.

Incidentally, Google didn't use the politically incorrect two letter acronym in the blog post, but they put it, with periods (?), in the post title.

No comments: