Over-Fitting Mitigation
One of the most successful strategies for mitigating over fitting in neural nets is to train different sets of neurons on different data sets. A software developer may divide a neural net into multiple sets of neurons. The developer then turns off all but one set of neurons and trains that set with a first batch of training data. He then repeats the process with the other neurons sets, training each neuron set with a different batch of training data with the other sets are turned off. Then all the neuron sets are turned on when data is analyzed.
So trained, the neural net acts like multiple brains with different backgrounds and experiences. The neural net is just like a group of people with diverse backgrounds. You can use a similar strategy with cerebral sex.