Empirical Performance Comparison of Two Symbolic Learning Systems Based on Selective and Constructive Induction
Witold Szczepanik, Tomasz Arciszewski · 2009
The paper provides results of a performance comparison study of two symbolic learning programs, both based on the AQ15C learning algorithm. The first program uses the single representation space while the second one utilizes constructive induction learning, which incorporates changes in the representation space. The performance of the compared systems was analyzed using the overall empirical error rates determined using the leave-one-out and hold-out sampling methods. Both system's performance was calculated for individual stages in a multi-stage knowledge acquisition process, and learning curves and their envelopes were prepared. The study was conducted using the set of 384 optimal designs of wind bracing in steel skeleton structures of tall buildings. Key words: structural engineering, learning design rules, selective and constructive induction, performance comparison of learning systems, empirical error rates, learning curves and envelopes. 1 Introduction There is a growing intere...