neural-ode

The paper presents a controlled matched-integrator evaluation of Hamiltonian Neural Networks against a parameter-matched baseline on pendulum and Kepler dynamics, showing significant reductions in energy drift and trajectory error. It also examines the behavior of Störmer–Verlet-style rollouts with learned non-separable Hamiltonians.

This paper proposes a neural ODE-based regularization method that enforces latent embeddings in reinforcement learning agents to follow consistent ODE flows, aligning representation learning with environment dynamics and yielding performance gains on Atari and gridworld benchmarks.

Introduces HyperODE, a zero-shot surrogate that maps ODE structures to hypergraphs, enabling simulation and parameter inference across entire families of dynamical systems without retraining.

This paper introduces PCINN, a physics-chemistry-informed neural network that acts as a hybrid AI surrogate for real-time spatial atomic layer deposition (SALD) coverage prediction, achieving CFD-level accuracy in ~77ms and enabling reliable kinetics inversion via identifiability analysis.

This paper presents a novel approach combining Graph Neural Networks with augmented Neural Ordinary Differential Equations (GNODE) for stable and accurate spatio-temporal prediction of unsteady airfoil aerodynamics, outperforming autoregressive baselines on transonic shock and non-linear dynamics tests.

This paper introduces OAT, a lightweight failure attribution tool for LLM-based agentic systems that trains only on successful trajectories and uses neural controlled differential equations to detect error steps, outperforming expensive baselines by orders of magnitude in speed and accuracy.

This paper introduces GNOVA, a GRU-Neural ODE Variational Autoencoder framework for reconstructing and forecasting Alzheimer's disease cognitive trajectories from routine clinical data without expensive neuroimaging or biomarkers, achieving low error and uncertainty estimation on the ADNI dataset.

This paper introduces steerable neural ordinary differential equations on homogeneous spaces, providing a geometric framework for learning continuous-time equivariant dynamics.

FFJORD introduces a scalable reversible generative model using continuous dynamics and Hutchinson's trace estimator to enable unbiased log-density estimation without architectural constraints. The method achieves state-of-the-art results on density estimation and image generation while maintaining efficient sampling.