In neural networks, we can use either case updating or batch updating to adjust the parameters (link weights and biases) of the network structure. What is one of the reasons that we may prefer batch updating, even if we expect that case updating would be more accurate in terms of the final outcome?