CMS, which progressively merges multiple-shooting windows until the single-shooting objective is recovered, improves convergence and generalization for NODE, UDE, and mechanistic ODE training across twelve benchmarks.
Levine, Dessi P
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
q-bio.QM 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Curriculum Multiple Shooting for Robust Training of Neural and Universal Differential Equations
CMS, which progressively merges multiple-shooting windows until the single-shooting objective is recovered, improves convergence and generalization for NODE, UDE, and mechanistic ODE training across twelve benchmarks.