Modulation-based meta-learning in a Hamiltonian framework enables accurate few-shot adaptation and generalization across parameter space for structure-preserving dynamics without explicit system parameterization.
The bottom panels show the target energy function and the learned energy functions by using the modulation methods
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Meta-learning Structure-Preserving Dynamics
Modulation-based meta-learning in a Hamiltonian framework enables accurate few-shot adaptation and generalization across parameter space for structure-preserving dynamics without explicit system parameterization.