More Efficient Windowing

Windowing has been proposed as a procedure for efficient memory use in the ID3 decision tree learning algorithm. However, previous work has shown that windowing may often lead to a decrease in performance. In this work, we try to argue that rule learning algorithms are more appropriate for windowing than decision tree algorithms, because the former typically learn and evaluate rules independently and are thus less susceptible to changes in class distributions. Most importantly, we present a new windowing algorithm that achieves additional gains in efficiency by saving promising rules and removing examples covered by these rules from the learning window. While the presented algorithm is only suitable for redundant, noise-free data sets, we will also briefly discuss the problem of noisy data for windowing algorithms.

Authors

• Johannes Fürnkranz

Keywords

No keywords are indexed for this paper.