data.in

This input file defines how the network reads in and interprets the data. You can also decide on the sort of output you want to get after training is completed. Here is an example, taken from the kappa problem:

1 flag_ts
1 flag_cross
0 flag_prepro
1 flag_end_____0_none_____1_all_outputs_____2_condpro

Row 1
Flag variable toggling between interpolation (0) and time series prediction (>1). In the latter case, the positive integer number indicates how many time steps ahead a prediction is to be made.

Row 2
Flag variable deciding whether to do the training with (1) or without (0) cross-validation. If set to 1, the data for cross-validation need to be available in file cross.in.

Row 3
Flag variable for preprocessing of the data. If set to 1, the data are normalised.

Row 4
Flag variable for defining the output after training is completed. 0: No output. 1: For every data point, the input to each of the nodes in the kernel layer (that is, the kernel centres) is calculated. The results are written out to file predict.out. From this you can obtain a state-space plot for the kernel centres. 2: Predicting the conditional probability density P(y|x) for a given input vector x.


Last modified: 19 May 2000