An Empirical Investigation of Brute Force to choose Features, Smoothers and Function Approximators (1995)

Andrew Moore, Daniel Hill, Michael Johnson

Tags

Locally Weighted Learning, Memory-based Learning

Abstract

The generalization error of a function approximator, feature set or smoother can be estimated directly by the leave-one-out cross-validation error. For memory-based methods, this is computationally feasible. We describe an initial version of a general memory-based learning system (GMBL): a large collection of learners brought into a widely applicable machine-learning family. We present ongoing investigations into search algorithms which, given a dataset, find the family members and features that generalize best. We also describe GMBL's application to two noisy, difficult problems---predicting car engine emissions from pressure waves, and controlling a robot billiards player with redundant state variables.

Full text

Download (application/pdf, 453.4 kB)

Approximate BibTeX Entry

@incollection{moore-empirical,
    Year = {1995},
    Volume = {III: Selecting Good Models},
    Pages = {361--379},
    Publisher = {MIT Press},
    Booktitle = {Computational Learning Theory and Natural Learning},
    Editor = {S. Hanson and S. Judd and T. Petsche},
    Author = { Andrew Moore, Daniel Hill, Michael Johnson },
    Title = {An Empirical Investigation of Brute Force to choose Features, Smoothers and Function Approximators}
}