In this type of Neural Network neurons, weights and bias become dual quaternions instead of scalar values.