Created by: cactiball

I have a question about the regression_cost, I did not find enough information about it in the document.

So I have a dataset with 0-1 labels. 1 represent the same, 0 represent different. If I build the network like this:

output = tensor_layer(
    a=input1,
    b=input2,
    size=1,
    act=SigmoidActivation(),
)

cost = regression_cost(
        input=output,
        label=lbl,
)
outputs(cost)

Is the "output" here represent the probability of "input1 and input2 are the same"?

Thanks!