Weighted Decision Trees

: Whiledecision tree compilationis a promisingway tocarry out guard tests e ciently, the methods given in the literature do not take into account either the execution characteristics of the program or the machine-level tradeo s between di erent ways to implement branches. These methods therefore o er little or no guidance for the implementorwith regard to how decision trees are to be realized on a particular machine. In this paper, we describe an approach that takes execution frequencies of di erent program branches, as well as the costs of alternative branch realizations, to generate decision trees. Experiments indicate that the performance of our approach is uniformly better than that of plausible alternative schemes.

• Type: Conference
• Year: 1992
• Where: ICLP
• Authors: Saumya K. Debray, Sampath Kannan, Mukul Paithane
• Keywords: Alternative Branch Realizations, Automated Reasoning, Decision Trees, Di Erent Ways, ICLP 1992