Random feature models for learning interacting dynamical systems

Random feature models for learning interacting dynamical systems
复制标题

用于学习交互动力系统的随机特征模型

DOI:
10.1098/rspa.2022.0835
复制
发表时间:
2023
期刊:
Physical and Engineering Sciences
影响因子:
--
通讯作者:
Schaeffer, Hayden
Schaeffer, Hayden
中科院分区:
--
文献类型:
--
作者:
Liu, Yuxuan;McCalla, Scott G.;Schaeffer, Hayden

文献摘要

参考文献

被引文献

相似文献

粒子动力学和多智能体系统为研究和预测复杂相互作用系统的行为提供了精确的动力学模型。它们通常采用高维微分方程系统的形式,由相互作用内核参数化,该内核模拟了代理之间潜在的吸引力或排斥力。我们考虑的问题是,直接从智能体在时间上的路径的噪声观测中构造一个基于数据的相互作用力的近似。然后使用学习到的交互核来预测代理在较长时间间隔内的行为。在这项工作中开发的近似使用随机特征算法和稀疏随机特征方法。稀疏性促进回归提供了一种机制来修剪随机生成的特征,当数据有限时,这种机制被观察到是有益的,特别是导致比其他方法更少的过拟合。此外,施加稀疏性降低了核评估成本,从而显著降低了多智能体系统预测的仿真成本。我们的方法应用于各种例子,包括具有齐次和非均匀相互作用的一阶系统,二阶齐次系统和一个新的羊群系统。
Particle dynamics and multi-agent systems provide accurate dynamical models for studying and forecasting the behaviour of complex interacting systems. They often take the form of a high-dimensional system of differential equations parameterized by an interaction kernel that models the underlying attractive or repulsive forces between agents. We consider the problem of constructing a data-based approximation of the interacting forces directly from noisy observations of the paths of the agents in time. The learned interaction kernels are then used to predict the agents’ behaviour over a longer time interval. The approximation developed in this work uses a randomized feature algorithm and a sparse randomized feature approach. Sparsity-promoting regression provides a mechanism for pruning the randomly generated features which was observed to be beneficial when one has limited data, in particular, leading to less overfitting than other approaches. In addition, imposing sparsity reduces the kernel evaluation cost which significantly lowers the simulation cost for forecasting the multi-agent systems. Our method is applied to various examples, including first-order systems with homogeneous and heterogeneous interactions, second-order homogeneous systems, and a new sheep swarming system.
从高度损坏的数据中精确恢复混沌系统
DOI: 10.1137/16m1086637
发表时间: 2016
期刊: Multiscale Model. Simul.
影响因子: --
作者:
Giang Tran;Rachel A. Ward
通讯作者: Rachel A. Ward
DOI: 10.48550/arxiv.2203.03979
发表时间: 2022-03
期刊: Proceedings of machine learning research
影响因子: --
作者:
D. Messenger;E. Dall’Anese;D. Bortz
通讯作者: D. Messenger;E. Dall’Anese;D. Bortz
DOI: 10.1016/j.acha.2022.08.003
发表时间: 2021-03
影响因子: 2.5
作者:
Abolfazl Hashemi;Hayden Schaeffer;Robert Shi;U. Topcu;Giang Tran;Rachel A. Ward
通讯作者: Abolfazl Hashemi;Hayden Schaeffer;Robert Shi;U. Topcu;Giang Tran;Rachel A. Ward
DOI: 10.1016/j.spa.2020.10.005
发表时间: 2019-12
影响因子: 1.4
作者:
Zhongyan Li;F. Lu;M. Maggioni;Sui Tang;C. Zhang
通讯作者: Zhongyan Li;F. Lu;M. Maggioni;Sui Tang;C. Zhang
基于稀疏辨识的非线性二时间尺度过程模型预测控制
DOI: 10.1016/j.compchemeng.2021.107411
发表时间: 2021
期刊: Comput. Chem. Eng.
影响因子: --
作者:
Fahim Abdullah;Zhe Wu;P. Christofides
通讯作者: P. Christofides