AURA: Angular Update Rate Adaptation for training complex-valued neural networks (Q7353577)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
No description defined
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | AURA: Angular Update Rate Adaptation for training complex-valued neural networks |
No description defined |
Statements
AURA: Angular Update Rate Adaptation for training complex-valued neural networks (English)
0 references
22 September 2026
0 references
cs.LG
0 references
math.NA
0 references
Enrico Ballini
0 references
Allan Peter Engsig-Karup
0 references
Tito Andriollo
0 references
AURA is a per-parameter step-size adaptation for complex-valued neural networks that measures agreement between consecutive updates to enlarge or reduce the update step size based on consistency, alignment, and rotation of complex parameters; AURA improves convergence with small overhead compared to Adam, Muon, and SGD on various test cases.
0 references
Step-size Adaptation
0 references
Complex-valued Neural Networks
0 references
Learning Rate Adaptation
0 references
1 reference
1 reference
1 reference
1 reference
1 reference
1 reference
1 reference
1 reference
1 reference
