Hi,
what is the activation function of the output neuron ?
The output column of your training set is outside the range [-1..1].
Maybe that is part of the problem.
Regards
Thorsten
Lena Schüssler wrote:
I have a small NN library for multilayer perceptron. It works fine on
classification problems, but when i try to train it on one simple data
set with 2 inputs in order to approximate function sin(a)+cos(b) the
error doesn't reduce.
If somebody of you could try to train NN on this dataset and tell me
if this task is appropriate for such type of networks or not i would
appreciate it
I just want to know if something is wrong with my network as i
programmed it myself or if this task is not for the network.
So here is the dataset
3.140000 0.015700 1.001469
4.580000 0.015700 0.008627
6.020000 0.015700 0.739719
4.430000 0.015700 0.039484
3.060000 0.015700 1.081379
0.640000 0.015700 1.597072
4.480000 0.015700 0.026758
5.470000 0.015700 0.273397
2.150000 0.015700 1.836776
0.430000 0.015700 1.416748
4.360000 0.015700 0.061326
4.800000 0.015700 0.003712
4.870000 0.015700 0.012272
1.200000 0.015700 1.931916
1.850000 0.015700 1.961152
1.870000 0.015700 1.955448
2.080000 0.015700 1.873010
1.780000 0.015700 1.978073
2.620000 0.015700 1.498139
0.650000 0.015700 1.605063
0.140000 0.015700 1.139420
0.650000 0.015700 1.605063
2.260000 0.015700 1.771629
5.050000 0.015700 0.056328
4.930000 0.015700 0.023461
1.730000 0.015700 1.987231
0.420000 0.015700 1.407637
5.220000 0.015700 0.125968
4.370000 0.015700 0.057922
2.610000 0.015700 1.506784
3.190000 0.015700 0.951488
1.230000 0.015700 1.942366
0.910000 0.015700 1.789381
2.930000 0.015700 1.209894
5.980000 0.015700 0.701315
3.970000 0.015700 0.263021
3.900000 0.015700 0.312111
4.180000 0.015700 0.138280
3.480000 0.015700 0.667892
0.090000 0.015700 1.089755
4.610000 0.015700 0.005114
1.880000 0.015700 1.952453
5.210000 0.015700 0.121151
3.520000 0.015700 0.630436
3.080000 0.015700 1.061431
0.780000 0.015700 1.703156
5.720000 0.015700 0.465994
5.170000 0.015700 0.102766
2.880000 0.015700 1.258496
2.380000 0.015700 1.689952
6.140000 0.031400 0.856810
3.030000 0.031400 1.110868
3.040000 0.031400 1.100925
2.120000 0.031400 1.852448
2.120000 0.031400 1.852448
2.010000 0.031400 1.904598
3.850000 0.031400 0.348882
2.870000 0.031400 1.267773
0.960000 0.031400 1.818699
2.260000 0.031400 1.771260
5.480000 0.031400 0.279935
4.470000 0.031400 0.028740
3.820000 0.031400 0.371953
0.430000 0.031400 1.416378
1.440000 0.031400 1.990965
3.840000 0.031400 0.356508
4.720000 0.031400 -0.000464
5.340000 0.031400 0.190074
1.750000 0.031400 1.983493
1.930000 0.031400 1.935684
5.430000 0.031400 0.246128
0.400000 0.031400 1.388925
3.810000 0.031400 0.379770
4.360000 0.031400 0.060956
4.250000 0.031400 0.104518
0.940000 0.031400 1.807065
5.460000 0.031400 0.266192
4.010000 0.031400 0.236206
6.110000 0.031400 0.827186
2.390000 0.031400 1.682310
0.110000 0.031400 1.109285
5.970000 0.031400 0.691416
5.740000 0.031400 0.482641
3.150000 0.031400 0.991100
2.130000 0.031400 1.847185
1.580000 0.031400 1.999465
5.490000 0.031400 0.286915
5.980000 0.031400 0.700945
4.450000 0.031400 0.033734
0.170000 0.031400 1.168689
2.290000 0.031400 1.751838
3.970000 0.031400 0.262651
4.640000 0.031400 0.002126
6.110000 0.031400 0.827186
4.720000 0.031400 -0.000464
6.080000 0.031400 0.797717
3.670000 0.031400 0.495348
3.170000 0.031400 0.971103
5.470000 0.031400 0.273027
5.740000 0.031400 0.482641
here 1-st column is "a" and second column is "b", third column is the
output.
Thank you and greetings,
leen- Skrij navedeno besedilo -
- Prikaži citirano besedilo -
