← Back to all news
arXiv Machine Learning By Yiheng (Roger), Jiang, Youssef Marzouk, Avishek Joey Bose, Aram-Alexandre Pooladian

Learning smooth dynamics via Kinematic Interpolants

Read the original on arXiv Machine Learning →

The paper introduces Kinematic Interpolants (KI), a new framework that extends stochastic interpolants to infer nonlinear dynamics from multiple snapshots in phase space, incorporating both position and velocity measurements. KI addresses the ambiguity in dynamics inference from real-world data and is scalable without simulation or unnecessary approximations. Experiments on various scientific datasets show that KI reconstructs underlying dynamics more accurately than existing baselines while requiring significantly less training time.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Machine Learning
Aug 5

Trajectory inference via Acceleration Matching

arXiv:2608. 03916v1 Announce Type: new Abstract: Trajectory inference is a fundamental problem in many scientific domains: given a collection of unpaired snapshots of observations at discrete time points, the goal is to generate smooth trajectories that best resemble and interpolate the data.

By Bartolo Dazzini, Giovanni Conforti, Alain Durmus, Aram-Alexandre Pooladian
benchmarks
More like this →
arXiv Machine Learning
Oct 5

Simulation-Free Learning of Population Dynamics with Wasserstein Lagrangian Residuals

The paper introduces Double‑Stitch, a simulation‑free method for learning population dynamics in Wasserstein space. It penalizes the residual of the equation of motion along a learned path, derived from a Clebsch variational principle that avoids gradient velocities. Experiments on synthetic, single‑cell, and ocean vortex data show that Double‑Stitch matches or surpasses gradient‑flow and simulation‑based methods while training 4–14 times faster.

By Fedor Sergeev, Markus Heinonen, Daniel Waxman, Tim Cooijmans, Ricardo Baptista, Dmitry Batenkov, Eli Bingham
More like this →
arXiv Machine Learning
Sep 2

Efficient Adaptation of ROMs for Unsteady Flows Using Data Assimilation

The paper presents a lightweight retraining strategy for a parameterized Reduced Order Model (ROM) that achieves full‑model accuracy using only a fraction of the computational effort and sparse observations. The ROM architecture combines a Variational Autoencoder for dimensionality reduction with a transformer network that evolves latent states while accounting for the Reynolds number as an external control variable. By leveraging the probabilistic VAE, the method generates trajectory ensembles and uncertainty estimates, and adapts to out‑of‑sample parameters through sparse data assimilation with an ensemble Kalman filter, focusing retraining on the autoencoder to correct latent manifold distortions.

By Isma\"el Zighed, Andrea N\'ovoa, Luca Magri, Taraneh Sayadi
llmsfine-tuning
More like this →