Lottery ticket hypothesis

In machine learning, the lottery ticket hypothesis is that artificial neural networks with random weights can contain subnetworks which entirely by chance can be tuned to a similar level of performance as the complete network.^[1] The term derived from considering the tunable subnetwork as the equivalent of a winning lottery ticket; the chance of any given ticket winning is tiny, but if you buy enough of them you are certain to win, and the number of possible subnetworks increases exponentially as the power set of the set of connections, making the number of possible subnetworks astronomical for any reasonably large network.

Malach et. al. have proved a stronger version of the hypothesis, which is that a sufficiently overparameterized untuned network will typically contain a subnetwork that is already an approximation to the given goal, even before tuning.^[2] A similar result has been proven for the special case of convolutional neural networks.^[3]

See also

• Grokking (machine learning)
• Pruning (artificial neural network)