Concept Discovery by Decision Table Decomposition and its Application in Neurophysiology

This chapter presents a “divide-and-conquer” data analysis method that, given a concept described by a decision table, develops its description in terms of intermediate concepts described by smaller and more manageable decision tables. The method is based on decision table decomposition, a machine learning approach that decomposes a given decision table into an equivalent hierarchy of decision tables. The decomposition aims to discover the decision tables that are overall less complex than the initial one, potentially easier to interpret, and introduce new and meaningful intermediate concepts. The chapter introduces the decomposition method and, through decomposition-based data analysis of two neurophysiological datasets, shows that the decomposition can discover physiologically meaningful concept hierarchies and construct interpretable decision tables which reveal relevant physiological principles.

[1]  Bruce G. Buchanan,et al.  Learning Intermediate Concepts in Constructing a Hierarchical Knowledge Base , 1985, IJCAI.

[2]  Bruce G. Buchanan,et al.  The MYCIN Experiments of the Stanford Heuristic Programming Project , 1985 .

[3]  Marko Bohanec,et al.  DEX: An Expert System Shell for Decision Support • , 1990 .

[4]  Donald Michie,et al.  Problem Decomposition and the Learning of Skills , 1995, ECML.

[5]  Alen D. Shapiro,et al.  Structured induction in expert systems , 1987 .

[6]  Bernhard Pfahringer,et al.  Controlling Constructive Induction in CIPF: An MDL Approach , 1994, ECML.

[7]  John W. Clark,et al.  A distributed-parameter model of the myelinated nerve fiber. , 1991 .

[8]  Alan W. Biermann,et al.  Signature Table Systems and Learning , 1982, IEEE Transactions on Systems, Man, and Cybernetics.

[9]  R. Michalski.  Understanding the Nature of Learning: Issues and Research Directions , 1985 .

[10]  Tadeusz Luba,et al.  Decomposition of multiple-valued functions , 1995, Proceedings 25th International Symposium on Multiple-Valued Logic.

[11]  A. L. Samuel,et al.  Some studies in machine learning using the game of checkers. II: recent progress , 1967 .

[12]  Larry A. Rendell,et al.  Lookahead Feature Construction for Learning Hard Concepts , 1993, International Conference on Machine Learning.

[13]  Ivan Bratko,et al.  Constructing Intermediate Concepts by Decomposition of Real Functions , 1997, ECML.

[14]  Ivan Bratko,et al.  Machine Learning by Function Decomposition , 1997, ICML.

[15]  J. Wolpaw,et al.  Operantly conditioned motoneuron plasticity: possible role of sodium channels. , 1995, Journal of neurophysiology.

[16]  John R. Anderson,et al.  MACHINE LEARNING An Artificial Intelligence Approach , 2009 .