A local, unsupervised pruning rule based on activity fluctuations preserves task performance in trained recurrent networks and outperforms magnitude and second-order pruning baselines, with optimal rescaling weaker than theory predicts.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
q-bio.NC 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Effective pruning of task-trained recurrent neural networks using noisy fluctuations and connection rescaling
A local, unsupervised pruning rule based on activity fluctuations preserves task performance in trained recurrent networks and outperforms magnitude and second-order pruning baselines, with optimal rescaling weaker than theory predicts.