A new forward-pass algorithm computes exact n-th derivatives of deep networks in time linear in network size, avoiding repeated autodifferentiation.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Quasilinear Algorithm for Computing Higher-Order Derivatives of Deep Feed-Forward Neural Networks
A new forward-pass algorithm computes exact n-th derivatives of deep networks in time linear in network size, avoiding repeated autodifferentiation.