LBYL restores pruned CNNs by spreading each pruned filter's contribution across multiple preserved filters via a closed-form delivery matrix, outperforming one-to-one neuron merging without data or fine-tuning.
Pruning neural networks without any data by iteratively conserving synaptic flow,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Training-Free Restoration of Pruned Neural Networks
LBYL restores pruned CNNs by spreading each pruned filter's contribution across multiple preserved filters via a closed-form delivery matrix, outperforming one-to-one neuron merging without data or fine-tuning.