The paper applies pseudo-arclength continuation, a classical numerical method, to neural network optimization, claiming improved generalization over ADAM on small MNIST tasks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Principled Curriculum Learning using Parameter Continuation Methods
The paper applies pseudo-arclength continuation, a classical numerical method, to neural network optimization, claiming improved generalization over ADAM on small MNIST tasks.