why omit the bias in softmax regression

[RalphMao]

I am also wondering about it

[nik-ko]

"Another property that could potentially be interesting is that the softmax output is invariant under additive changes of its input vector: softmax(a) = softmax(a + b) when b is a scalar added to all the elements of vector a. " from: http://www.iro.umontreal.ca/~bengioy/dlbook/mlp.html#pf8

Or: http://ufldl.stanford.edu/wiki/index.php/Softmax_Regression