Page 308 -
P. 308

Appendix A. CD Datasets   297

                                    FTW  Foetal weight at birth
                                    FTL    Foetal length at birth
                                    CPB    Cephalic perimeter at birth

                                  Source: A Matos, S. Jog0 Hospital, Oporto.
                                  Reference: F Sereno et al. (2000) The Application of Radial Basis Functions and
                                  Support Vector Machines to the Foetal Weight Prediction. Intel1 Eng Syst Through
                                  Artif Neural Networks, 10: 801 -806.



                                  A.ll  Food

                                  The Food.xls file contains a table with the following contents for 36 food products:

                                    CAL          Calories per g
                                    P            Phosphor (mg per g)
                                    Ca           Calcium (mg per g)
                                    Fe           Iron  (mg per g)
                                    PROT         Proteic substances  (mg per g)
                                    BIB2         Vitamins B 1 and B2  (mg per g)

                                  Source: Portuguese pharmacy flyer, 1990.



                                  A.12  Fruits

                                  The Fruits  dataset contains colour image files of  apples (22),  oranges  (15) and
                                  peaches (36). The fruit images shown in chapter 1 (Figures 1.1 and  1.2) are also
                                  included. All  images are  in  256 colours, obtained  with  a  low-resolution digital
                                  camera.

                                  Source: JP Marques de SB, Engineering Faculty, Oporto University.


                                   A.13 Impulses on Noise


                                   The  Signal  Noise.xls  file  contains  100  samples  of  noise  with  a  chi-square
                                   distribution, to which were added impulses with arrival times following a Poisson
                                   distribution. The amplitudes of the impulses were also generated with a chi-square
                                   distribution.
                                     A  threshold  value  can  be  specified  in  order  to  detect  the  signal  impulses.
                                   Changing  the  value  of  the  threshold  will  change  the  number of  true  and  false
                                   impulse detections.
   303   304   305   306   307   308   309   310   311   312   313