A method to construct a steering vector by subtracting the mean activation in one class of examples from the mean of another.
A method to construct a steering vector by subtracting the mean activation in one class of examples from the mean of another.