Margin Classifier - Generalization Error Bounds

Generalization Error Bounds

One theoretical motivation behind margin classifiers is that their generalization error may be bound by parameters of the algorithm and a margin term. An example of such a bound is for the AdaBoost algorithm. Let be a set of examples sampled independently at random from a distribution . Assume the VC-dimension of the underlying base classifier is and . Then with probability we have the bound

for all .

Read more about this topic:  Margin Classifier

Famous quotes containing the words error and/or bounds:

    “Still we live meanly, like ants; though the fable tells us that we were long ago changed into men; like pygmies we fight with cranes; it is error upon error, and clout upon clout, and our best virtue has for its occasion a superfluous and evitable wretchedness. Our life is frittered away by detail.”
    —Henry David Thoreau (1817–1862)

    “Prohibition will work great injury to the cause of temperance. It is a species of intemperance within itself, for it goes beyond the bounds of reason in that it attempts to control a man’s appetite by legislation, and makes a crime out of things that are not crimes. A Prohibition law strikes a blow at the very principles upon which our government was founded.”
    —Abraham Lincoln (1809–1865)