Favstar gets even better if you sign in.
Here's why signing in is good for you.
Awesome talk from Jeff Hawkins (@numenta) on machine intelligence. https://www.youtube.com/watch?v=4y43qwS8fl4&feature=player_embedded …
Deep Learning on EC2 spot instances. Sweet! pic.twitter.com/URQ9btKN
$./runexperiment m1.large ex.m --> boots m1.large EC2, runs ex.m, saves to S3 and shuts down. Setup: https://plus.google.com/u/1/106894657445385494273/posts/KJjViCg9Uny …
cc @gallamine
@ogrisel Yea. Good points. I've tried random projections (50-1000) + 3NN, and i can't get it past ~96.3%. Soft coding didn't do any good.
@ogrisel Hmm. btw, if PCA+3NN gives 98% then PCA + <cool method> is a bit moot since you could say it was just the PCA i guess.
@ogrisel Maybe random samples and soft encoding ( http://www.stanford.edu/~acoates/papers/coatesng_icml_2011.pdf …)
collectd-php class done. Can now push stats to collectd in my php apps. Awesome! http://t.co/gy6BNlx @auxesis
@flocto
Deep Learning, Machine Learning and Web tech. DeepLearnToolbox author.
Stats can't be shown as @rasmusbergpalm has never signed in to Favstar.
