Partial fusion of neural networks — merging only the most similar neurons via partial optimal transport or clustering — yields a monotone, tunable accuracy/parameter tradeoff between weight aggregation and ensembles.
arXiv preprint arXiv:2210.06671 , year =
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2representative citing papers
Hierarchical Re-Basin merging induces stronger adversarial and perturbation robustness in combined models as more participants are added, but produces larger performance drops than previously reported.
citing papers explorer
-
Partial Fusion of Neural Networks: Efficient Tradeoffs Between Ensembles and Weight Aggregation
Partial fusion of neural networks — merging only the most similar neurons via partial optimal transport or clustering — yields a monotone, tunable accuracy/parameter tradeoff between weight aggregation and ensembles.
-
Robustness and Regularization in Hierarchical Re-Basin
Hierarchical Re-Basin merging induces stronger adversarial and perturbation robustness in combined models as more participants are added, but produces larger performance drops than previously reported.