crobb305 - 6:44 pm on Apr 16, 2011 (gmt 0)
So how to run hundreds of different algos
You look for clustering and focus in on a final result. Outliers would get excluded. This is what happens in any environment that relies on multiple numerical models (algorithms), each using different equations, physics, parameterizations, etc. Over time, the machine (statistical algorithms) can learn to identify an expected result based on a statistical dataset (probability distribution) of previous events/cases/results or from tests using random variables.