BigML

I was just checking out this ML as a service.

https://bigml.com/pricing

On pricing, if you select the Terabyte bucket, and then push the sliders ALL THE WAY over to the right.. at 5TB + 5TB, and add up to a whopping $26K, we are not even close to the size of the data and models we are doing daily ... times 10… Probably not an economical solution. (I do like their blog though)

In this post: http://blog.bigml.com/2013/03/18/bedtime-for-boosting/ 

They mention: 

"The authors show that, if you carefully construct a dataset, you can make certain types of boosters (ones that optimize convex potential functions, for those who care) produce models that perform no better than chance by adding just a single carefully selected noise instance."

-JD