The authors prove that out-of-distribution convergence degradation in learning to optimize scales with the magnitude of the model's input feature shift, and build a gradient-only optimizer that reduces this shift.
A Deep Q-Network Based-Resource Allocation Scheme for Massive MIMO-NOMA
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Robust Learning to Optimize with Theoretical Guarantees
The authors prove that out-of-distribution convergence degradation in learning to optimize scales with the magnitude of the model's input feature shift, and build a gradient-only optimizer that reduces this shift.