Insights into the Lottery Ticket Hypothesis and the Iterative Magnitude Pruning

arXiv:2403.15022v1 Announce Type: new
Abstract: Lottery ticket hypothesis for deep neural networks emphasizes the importance of initialization used to re-train the sparser networks obtained using the iterative magnitude pruning process. An explanation for why the specific initialization proposed by the lottery ticket hypothesis tends to work better in terms of generalization (and training) performance has been lacking. Moreover, the underlying principles in iterative magnitude pruning, like the pruning of smaller magnitude weights and the role of the iterative process, lack …

abstract arxiv cs.lg hypothesis importance insights iterative lottery ticket hypothesis networks neural networks process pruning terms train type work