Inductive algorithms rely strongly on their representational biases. Constructive induction can mitigate representational inadequacies. This paper introduces the notion of a relative gain measure and describes a new constructive induction algorithm (GALA) which is independent of the learning algorithm. Unlike most previous research on constructive induction, our methods are designed as preprocessing step before standard machine learning algorithms are applied. We present the results which demonstrate the effectiveness of GALA on artificial and real domains for several learners: C4.5, CN2, perceptron and backpropagation.
|Number of pages||6|
|State||Published - 1 Dec 1996|
|Event||Proceedings of the 1996 13th National Conference on Artificial Intelligence, AAAI 96. Part 1 (of 2) - Portland, OR, USA|
Duration: 4 Aug 1996 → 8 Aug 1996
|Conference||Proceedings of the 1996 13th National Conference on Artificial Intelligence, AAAI 96. Part 1 (of 2)|
|City||Portland, OR, USA|
|Period||4/08/96 → 8/08/96|