This paper presents an empirical study of the use of the rough set app
roach to reduction of data for a neural network classifying objects de
scribed by quantitative and qualitative attributes. Two kinds of reduc
tion are considered: reduction of the set of attributes and reduction
of the domains of attributes. Computational tests were performed with
five data sets having different character, for original and two reduce
d representations of data. The learning time acceleration due to data
reduction is up to 4.72 times. The resulting increase of misclassifica
tion error does not exceed 11.06%. These promising results let us clai
m that the rough set approach is a useful tool for preprocessing of da
ta for neural networks.