M. Kubat et D. Flotzinger, TREE-STRUCTURES OF LINEAR THRESHOLD UNITS FOR THE CLASSIFICATION OF NUMERIC EXAMPLES, Cybernetics and systems, 26(5), 1995, pp. 521-533
A decision tree-based system for learning from numeric data is describ
ed. Results from linear algebra (pseudoinverse matrices) help the syst
em to generate decision trees where the nodes are represented by linea
r threshold units minimizing the mean square error. The system's capab
ility to provide good classifications with small decision trees is dem
onstrated on artificial and benchmark data.