Revisiting Loss Modelling for Unstructured Pruning
This work addresses the problem of designing effective pruning criteria for reducing computational costs in neural networks, but it is incremental as it refines existing loss modeling approaches.
The paper revisits loss modeling for unstructured pruning in deep neural networks, showing that both first and second order Taylor expansions achieve similar performance, and that preserving network function does not always lead to better post-fine-tuning results.
By removing parameters from deep neural networks, unstructured pruning methods aim at cutting down memory footprint and computational cost, while maintaining prediction accuracy. In order to tackle this otherwise intractable problem, many of these methods model the loss landscape using first or second order Taylor expansions to identify which parameters can be discarded. We revisit loss modelling for unstructured pruning: we show the importance of ensuring locality of the pruning steps. We systematically compare first and second order Taylor expansions and empirically show that both can reach similar levels of performance. Finally, we show that better preserving the original network function does not necessarily transfer to better performing networks after fine-tuning, suggesting that only considering the impact of pruning on the loss might not be a sufficient objective to design good pruning criteria.