Patrick Cloutier, Cristian Tibirna, Bernard Grandjean et Jules Thibault
Département de génie chimique Université Laval
Sainte-Foy (Québec) CANADA G1K 7P4
NNFit (Neural Network Fitting) is a non-linear regression program based on multilayered neural network models. NNFit allows the userfriendly development and application of neural emprirical relations between input and output variables. For the sake of usability, the program displays:
· various tables that give statistical informations on prediction model errors (minimum, maximum, mean, standard deviation, correlation coefficients)
· various figures that easier to judge the performance of the developped models.
Copies of the documentation are available in many forms: HTML on line, HTML pages archive, PostScript file archive. If problems arise for downloading these ressources, we can also send the documentation by e-mail. Send a message having the subjet "Query for documentation" and indicating the e-mail address to which the send to be done, and also the format (HTML or PostScript).
The program is available for the platforms indicated here after, as precompiled binaries. Each distribution file contains the executable and a data file demo.dat (the examples illustrated in the manual are done with these data). The binaries should function at least on systems compatible with the ones indicated in the table.
The XForms library is