should a neural network input have only one target group?
1 view (last 30 days)
hai i have a doubt in neural network. i have 3 columns of input vector for which the target value is from 1 to 3 for each row. As shown below if the row vector is 2 4 and 7 the output should be 1. if it is 4 7 8 the output should be 2. neural network worked fine with this type of data.
2 4 7 1
4 7 8 2
2 5 9 3
now i have another set of data for which i have two row vectors with the same target value.
2 4 7 1
6 7 2 1
with the same target 1. the same will continue for the whole row vectors sometimes there is same 3 target values for different row vectors. should i make some modification in the input data? will the neural network work for this type of data. i tried but i dont get near exact values. please help in this regard
Greg Heath on 2 Apr 2017
All that can be guaranteed (if you design enough nets with different random initial weights and numbers of hidden nodes) is that the mse will be minimized to a reasonable degree of accuracy.
Hope this helps,
Thank you for formally accepting my answer