Integrated Model — A Proposal to handle noise
We present the theoretical background of the Integrated Model, a new induction algorithm proposed and implemented by the authors. The algorithm relies on a bottom-up strategy, from particular to general, feature less common that the usual top-down strategy founded in a great number of induction tools. We introduce a method for finding the values of the basic probability assignment according to Theory of Evidence, called probabilistic mass. With this notion, we propose a generalisation of the algorithm capable of handle noisy data.
Keywordsclassification rules decision trees generalization induction machine learning noise theory of evidence uncertainty measures
Unable to display preview. Download preview PDF.
- BREIMAN, L. et al. (1984) Classification and Regression Trees. Wadsworth & Brooks / Cole Advanced Books & Software. Pacific Grove, California.Google Scholar
- CLARK, P., (1989). Functional Specification of CN and AQ. (Reference: TI/P2154/PC/4/1.2). The Turing Institute.Google Scholar
- MOURA PIRES, F., (1993). Aprendizagem por Indu©Ão Empírica. Ph.D Thesis in Robotics. FCT-UNL, Lisboa, 1993.Google Scholar
- MOURA PIRES, F., NETO, J.P., (1995). Integrated Model — a new algorithm in Empirical Induction. Paper submitted to CAEPIA'95, VI Conference of the Spanish Association for Artificial Intelligence.Google Scholar
- NETO, J.P. (1995). Aprendizagem por Indu©Ão — Modelo Integrado. Mh. Thesis in Informatics. FCT-UNL, Lisboa, 1995.Google Scholar
- QUINLAN, J. R. (1993). C4.5: Programs for Machine Learning. Morgan Kaufman Publishers. San Mateo, California.Google Scholar
- SHAFER, G., (1976). A Mathematical Theory of Evidence. Princeton Univ. Press.Google Scholar