Conditional Flow Matching for Time Series Modelling

Ella Tamir, Najwa Laabid, Markus Heinonen, Vikas Garg, Arno Solin

Keywords: Conditional Flow Matching, Neural ODEs, Time-Series Modelling

TL;DR: We use Conditional Flow Matching to learn complex ODEs by regressing over conditional vector fields.

Abstract:

Learning dynamical systems from long trajectories is a challenging problem due to the complexity of the loss landscape.
Inspired by conditional flow matching in generative modelling,
we propose a new approach for training neural ODEs based on regressing vector fields of conditional probability paths
defined per trajectory.
Our Conditional Flow Matching for Time Series (CFM-TS) objective outperforms neural ODEs trained with the adjoint method
on three simulated tasks, including a pendulum system where the neural ODE does not converge at all.

Submission Number: 88