Searching Continuous State Spaces Effectively Using Critical Points TITLE2
Marc S. Atkin, Peter D. A. Cohen · 1999
Having access to massive amounts of data does not necessarily imply that induction algorithms must use them all. /textit{Samples} often provide the same accuracy with far less computational cost. However, the correct sample size is rarely obvious. We analyze methods for /textit{progressive sampling}---starting with small samples and progressively increasing them as long as model accuracy improves. We show that a simple geometric sampling schedule is in an asymptotic sense. We then explore the notion of optimal efficiency: what is the absolute best sampling schedule? We describe the issues involved in instantiating an optimaly efficient progressive sampler. Finally, we provide empirical results comparing a variety of progressive smampling methods. We conclude that progressive sampling often is preferable to analyzing all data instance.