ÍâÍøÌìÌÃ

ÌýNeuralÌýnetworksÌýareÌýoften studied analytically throughÌýscalingÌýlimits: regimes in which taking to infinityÌýÌýstructural network parameters such as depth, width, and numberÌýofÌýtraining datapoints results in simplified modelsÌýof learning. I will survey several such approaches with the goal of illustrating the rich and still not fully understood space of possible behaviors when some or all of the network’s structural parameters are large.