Matching Methods with Problems: A Comparative Analysis of Constructive Induction Approaches

This paper provides a taxonomy of constructive induction problems and reports on an empirical comparison of several constructive induction methods. In this paper a representation space is said to be poorly suited for learning because of three types of problems: 1) inappropriate attributes or attribute values, 2) incomplete attribute values, attribute sets or examples and 3) incorrect attributes or examples. Most current constructive induction methods are designed to correct one of these types (or sub-types) of problems which limits the types of problems for which this method is effective. In order to build a more general multistrategy method of constructive induction an understanding of when some methods for constructive induction are useful and when they fail is important. Five methods of constructive induction are evaluated: DCI attribute construction (AQDCI), HCI attribute construction (AQ-HCI(ADD)), HCI attribute removal (AQ-HCI(REMOVE)), HCI construction and removal (AQ-HCI), and attribute-value removal (AQ-SCALE). The results point to the need for a multistrategy constructive induction approach for solving a wide variety of induction problems.

[1]  Ryszard S. Michalski,et al.  CONSTRUCTIVE INDUCTION FROM DATA IN AQ17-DCI: Further Experiments , 1991 .

[2]  Larry A. Rendell,et al.  Learning hard concepts through constructive induction: framework and rationale , 1990, Comput. Intell..

[3]  Gheorghe Tecuci,et al.  Consistency-driven Knowledge Elicitation: Using a Learning-oriented Knowledge Representation that Su , 1993 .

[4]  Stefan Wrobel.  Demand-Driven Concept Formation , 1987, Knowledge Representation and Organization in Machine Learning.

[5]  Carla E. Brodley,et al.  Addressing the Selective Superiority Problem: Automatic Algorithm/Model Class Selection , 1993 .

[6]  Randy Kerber,et al.  ChiMerge: Discretization of Numeric Attributes , 1992, AAAI.

[7]  Ryszard S. Michalski,et al.  A theory and methodology of inductive learning , 1993 .

[8]  R. Michalski,et al.  Multistrategy Constructive Induction: AQ17-MCI , 1993 .

[9]  Ryszard S. Michalski,et al.  Discovering Representation Space Transformations for Learning Concept Descriptions Combining DNF and M-of-N Rules , 1994 .

[10]  Jerzy W. Bala,et al.  Iterative rule simplification for noise tolerant inductive learning , 1992, Proceedings Fourth International Conference on Tools with Artificial Intelligence TAI '92.

[11]  Katharina Morik,et al.  Knowledge Representation and Organization in Machine Learning , 1989, Lecture Notes in Computer Science.

[12]  Larry A. Rendell,et al.  Constructive Induction On Decision Trees , 1989, IJCAI.

[13]  David Tcheng,et al.  MORE ROBUST CONCEPT LEARNING USING DYNAMICALLY – VARIABLE BIAS , 1987 .

[14]  Janusz Wnek,et al.  Hypothesis-driven constructive induction , 1993 .

[15]  Nada Lavrac,et al.  The Multi-Purpose Incremental Learning System AQ15 and Its Testing Application to Three Medical Domains , 1986, AAAI.

[16]  George Drastal,et al.  Induction in an Abstraction Space: A Form of Constructive Induction , 1989, IJCAI.