Normal normalization with one

Question

I have a question about normalizing my dataset. We are working on a school assignment where we need to understand the data set and classify new examples. We have several datasets available that are compressed forms of the original. We tried to work with the smallest dataset, just to gain access to ANN.

A dataset consists of 8 columns of data and one for ideal values. The data columns are floating point values, and ideal values ​​are integers. The ideal field is 1 if the string belongs to the class and 0 does not. But when normalize () is applied to AnalystNormalizeCSV, the ideal field is converted to two fields.

Now suppose a simple direct neural network. Do I need one or two output neurons?

When I use 1 neuron and 1 for the number of ideal fields, it seems to work, but it hangs about 60%. When I use 2 output neurons and 1 for the number of ideal fields, I get an ArrayOutOfBoundsException in Propagation.iteration (). And when we use 2 for the number of output neurons and ideal fields, it works, but again hangs at 60%. The middle option seems normal, because in reality there is 1 ideal field, and after normalization there are 2 ideal fields, so there are 2 output neurons.

Default

Thanks in advance, Chris

Answer 1

If you use "one" normalization, you need two. You have two classes. You can model it with only one output neuron, but then it is much more regression (predict a number) than classification (which class).