When machines classify data, the rules can be arbitrarily complex. Humans, on the other hand, have trouble manipulating large structures and algebraic formulae. Therefore, when the rules are learned for the purpose of their execution by humans, the classier's accuracy needs to be balanced against the hardness of evaluating the rules given an unseen instance. This work examines several known learners in this light, and proposes a new algorithm for building human-friendly rules from datasets where easy classications exist. We also propose a new accuracy-complexity measure. In contrast to measures like BIC where the complexity penalty is rigid, ours allows exibility in the weight given to complexity. We evaluate our rule of thumb learner against the known learners, and show that it produces better results over a wide range of the complexity-accuracy trade-off parameter.
By: Dan Pelleg
Published in: H-0255 in 2006
LIMITED DISTRIBUTION NOTICE:
This Research Report is available. This report has been submitted for publication outside of IBM and will probably be copyrighted if accepted for publication. It has been issued as a Research Report for early dissemination of its contents. In view of the transfer of copyright to the outside publisher, its distribution outside of IBM prior to publication should be limited to peer communications and specific requests. After outside publication, requests should be filled only by reprints or legally obtained copies of the article (e.g., payment of royalties). I have read and understand this notice and am a member of the scientific community outside or inside of IBM seeking a single copy only.
Questions about this service can be mailed to reports@us.ibm.com .