WeightFlow: Learning Stochastic Dynamics via Evolving Weight of Neural Network
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908683165761536 |
|---|---|
| author | Li, Ruikun Liu, Jiazhen Wang, Huandong Liao, Qingmin Li, Yong |
| author_facet | Li, Ruikun Liu, Jiazhen Wang, Huandong Liao, Qingmin Li, Yong |
| contents | Modeling stochastic dynamics from discrete observations is a key interdisciplinary challenge. Existing methods often fail to estimate the continuous evolution of probability densities from trajectories or face the curse of dimensionality. To address these limitations, we presents a novel paradigm: modeling dynamics directly in the weight space of a neural network by projecting the evolving probability distribution. We first theoretically establish the connection between dynamic optimal transport in measure space and an equivalent energy functional in weight space. Subsequently, we design WeightFlow, which constructs the neural network weights into a graph and learns its evolution via a graph controlled differential equation. Experiments on interdisciplinary datasets demonstrate that WeightFlow improves performance by an average of 43.02\% over state-of-the-art methods, providing an effective and scalable solution for modeling high-dimensional stochastic dynamics. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_00451 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | WeightFlow: Learning Stochastic Dynamics via Evolving Weight of Neural Network Li, Ruikun Liu, Jiazhen Wang, Huandong Liao, Qingmin Li, Yong Computational Engineering, Finance, and Science Modeling stochastic dynamics from discrete observations is a key interdisciplinary challenge. Existing methods often fail to estimate the continuous evolution of probability densities from trajectories or face the curse of dimensionality. To address these limitations, we presents a novel paradigm: modeling dynamics directly in the weight space of a neural network by projecting the evolving probability distribution. We first theoretically establish the connection between dynamic optimal transport in measure space and an equivalent energy functional in weight space. Subsequently, we design WeightFlow, which constructs the neural network weights into a graph and learns its evolution via a graph controlled differential equation. Experiments on interdisciplinary datasets demonstrate that WeightFlow improves performance by an average of 43.02\% over state-of-the-art methods, providing an effective and scalable solution for modeling high-dimensional stochastic dynamics. |
| title | WeightFlow: Learning Stochastic Dynamics via Evolving Weight of Neural Network |
| topic | Computational Engineering, Finance, and Science |
| url | https://arxiv.org/abs/2508.00451 |