Automated Discovery of Empirical Laws

We define the problem of empirical search for knowledge by interaction with a setup experiment, and we present a solution implemented in the FAHRENHEIT discovery system. FAHRENHEIT autonomously explores multi-dimensional empirical spaces of numerical parameters, making experiments, generalizing them into empirical equations, finding the scope of applications for each equation, and setting new discovery goals, until it reaches the empirically complete theory. It turns out that a small number of generic goals and a small number of data structures, when combined recursively, can lead to complex discovery processes and to the discovery of complex theories. We present FAHRENHEIT's knowledge representation and the ways in which the discovery mechanism interacts with the emerging knowledge. Brief descriptions of several real-world applications demonstrate the system's discovery potential.

[1]  Jieming Zhu,et al.  Automated Discovery in a Chemistry Laboratory , 1990, AAAI.

[2]  Jieming Zhu,et al.  Operational Definition Refinement: A Discovery Process , 1992, AAAI.

[3]  Pat Langley,et al.  A Robust Approach to Numeric Discovery , 1990, ML.

[4]  Wojciech Ziarko,et al.  Rough Sets and Knowledge Discovery: An Overview , 1993, RSKD.

[5]  Raúl E. Valdés-Pérez,et al.  Conjecturing Hidden Entities by Means of Simplicity and Conservation Laws: Machine Discovery in Chemistry , 1994, Artif. Intell..

[6]  R. Bharat Rao,et al.  Learning Engineering Models with the Minimum Description Length Principle , 1992, AAAI.

[7]  Herbert A. Simon,et al.  Scientific discovery: compulalional explorations of the creative process , 1987 .

[8]  J. M. Żytkow.  An Interpretation of a Concept in Science by a Set of Operational Procedures , 1982 .

[9]  Herbert A. Simon,et al.  The Processes of Scientific Discovery: The Strategy of Experimentation , 1988, Cogn. Sci..

[10]  Peter W. Pachowicz,et al.  Fusion Of Vision And Touch For Spatio-Temporal Reasoning In Learning Manipulation Tasks , 1990, Other Conferences.

[11]  Cullen Schaffer.  Bivariate Scientific Function Finding in a Sampled, Real-Data Testbed , 1993, Mach. Learn..

[12]  Jan M. Zytkow,et al.  Discovery of Equations: Experimental Evaluation of Convergence , 1992, AAAI.

[13]  B W Koehn,et al.  Experimenting and theorizing in theory formation , 1986, ISMIS '86.

[14]  Henry S. Leonard,et al.  Testability and Meaning. , 1937 .

[15]  Marjorie Moulet.  A Symbolic Algorithm for Computing Coefficients' Accuracy in Regression , 1992, ML.

[16]  Herbert A. Simon,et al.  The Right Representation for Discovery: Finding the Conservation of Momentum , 1992, ML.

[17]  Pat Langley,et al.  An Integrated Approach to Empirical Discovery , 1989 .

[18]  Donald Gerwin,et al.  Information processing, data inferences, and scientific generalization , 1974 .

[19]  Donald Rose.  Using Domain Knowledge to Aid Scientific Theory Revision , 1989, ML.

[20]  Percy Williams Bridgman,et al.  The Logic of Modern Physics , 1927 .

[21]  Jan M. Zytkow,et al.  Scientific Model-Building as Search in Matrix Spaces , 1993, AAAI.