Predicting the Dynamics of Complex System via Multiscale Diffusion Autoencoder

TL;DR

This paper introduces MDPNet, a multiscale diffusion autoencoder-based model that captures the intrinsic dynamics of complex systems across multiple scales, significantly improving prediction accuracy and robustness.

Contribution

The paper presents a novel multiscale diffusion autoencoder combined with an attention-based graph neural ODE to better model multiscale dynamics in complex systems.

Findings

Achieves 53.23% reduction in prediction error compared to baselines.

Demonstrates superior robustness and generalization in complex system predictions.

Effectively captures multiscale structures for accurate spatiotemporal evolution modeling.

Abstract

Predicting the dynamics of complex systems is crucial for various scientific and engineering applications. The accuracy of predictions depends on the model's ability to capture the intrinsic dynamics. While existing methods capture key dynamics by encoding a low-dimensional latent space, they overlook the inherent multiscale structure of complex systems, making it difficult to accurately predict complex spatiotemporal evolution. Therefore, we propose a Multiscale Diffusion Prediction Network (MDPNet) that leverages the multiscale structure of complex systems to discover the latent space of intrinsic dynamics. First, we encode multiscale features through a multiscale diffusion autoencoder to guide the diffusion model for reliable reconstruction. Then, we introduce an attention-based graph neural ordinary differential equation to model the co-evolution across different scales. Extensive evaluations on representative systems demonstrate that the proposed method achieves an average prediction error reduction of 53.23% compared to baselines, while also exhibiting superior robustness and generalization.