with an artificial neural network, the weights are added to the connections, but the bias is added to the neuron. why does the network need to be set up this way?
A. Improved convergence
B. Enhanced generalization
C. Avoiding overfitting
D. Faster computation
