A branched signature kernel method with count-sampling turns single-trajectory rough signals into hierarchical paths for kernel-collocation ODE solving, supported by a universal approximation theorem and streaming updates.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.NA 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Branched Signature Kernel Solvers for ODEs with rough Single-Trajectory signals
A branched signature kernel method with count-sampling turns single-trajectory rough signals into hierarchical paths for kernel-collocation ODE solving, supported by a universal approximation theorem and streaming updates.