The Lottery Ticket Hypothesis was proposed by researchers at MIT in March 2018, and the relevant research results were published in the paper "..."The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks ".

The lottery hypothesis proposes that a randomly initialized dense neural network contains a subnetwork initialized in such a way that, when trained alone, it can achieve the same test accuracy as the original network within at most the same number of iterations. As a novel perspective on neural network composition, the lottery hypothesis enhances the academic community's theoretical understanding of neural networks.