COMBINATIONS OF WEAK CLASSIFIERS

To obtain classification systems with both good generalization perform ance and efficiency in space and time, we propose a learning method ba sed on combinations of weak classifiers, where weak classifiers are li near classifiers (perceptrons) which can do a little better than makin g random guesses. A randomized algorithm is proposed to find the weak classifiers. They are then combined through a majority vote. As demons trated through systematic experiments, the method developed is able to obtain combinations of weak classifiers with good generalization perf ormance and a fast training time on a variety of test problems and rea l applications. Theoretical analysis on one of the test problems inves tigated in our experiments provides insights on when and why the propo sed method works. In particular, when the strength of weak classifiers is properly chosen, combinations of weak classifiers can achieve a go od generalization performance with polynomial space- and time-complexi ty.