One-shot learning for the complex dynamical behaviors of weakly nonlinear forced oscillators

2026-04-16 · Source: Machine Learning · Field: Science & Research — Engineering & Applied Sciences, Mathematics & Computational Sciences · Depth: Expert, quick

Summary

A new one-shot learning method, MEv-SINDy (Multi-frequency Evolutionary Sparse Identification of Nonlinear Dynamics), has been developed to predict complex nonlinear dynamics from a single excitation time history. This method infers governing equations for non-autonomous and multi-frequency systems by leveraging the Generalized Harmonic Balance (GHB) method to decompose forced responses into slow-varying evolution equations. The study validated MEv-SINDy on two Micro-Electro-Mechanical Systems (MEMS): a nonlinear beam resonator and a MEMS micromirror. The model, trained on just one data point, accurately predicted softening/hardening effects and jump phenomena across various excitation levels, significantly reducing data acquisition needs for nonlinear microsystem characterization and design.

Key takeaway

For AI Scientists and Research Scientists working on complex nonlinear systems, MEv-SINDy offers a powerful approach to drastically reduce data requirements for system characterization. You should consider integrating this one-shot learning methodology to infer governing equations and predict dynamic behaviors like jump phenomena, especially in data-scarce environments such as MEMS design and validation.

Key insights

MEv-SINDy enables one-shot learning of complex nonlinear dynamics from minimal data.

Principles

• Infer governing equations from single time histories.
• Decompose complex responses into slow-varying evolution equations.

Method

MEv-SINDy combines sparse identification of nonlinear dynamics with Generalized Harmonic Balance to learn governing equations from a single excitation time history.

In practice

• Characterize MEMS with reduced data acquisition.
• Predict softening/hardening effects from one data point.

Topics

• One-shot Learning
• Nonlinear Dynamics
• MEv-SINDy
• Generalized Harmonic Balance
• MEMS

Best for: AI Scientist, Research Scientist

Related on AIssential

• How Low Can You Go? Active Learning for Sparse Model Discovery in the Ultra-Low-Data Limit — An active learning strategy using E-SINDy efficiently discovers sparse dynamical models by prioritizing informative data samples in ultra-low data regimes.
• Chaotic Oscillator Networks for Classification Tasks — Machine learning can approximate coupling terms in chaotic oscillator networks for scalable, efficient classification and pattern recognition.
• Integral Formulation of QENDy for Robust Nonlinear System Identification — Removing time derivative dependency significantly enhances nonlinear system identification robustness.
• An explicit operator explains end-to-end computation in the modern neural networks used for sequence and language modeling — SSMs compute by encoding inputs as traveling waves in a nonlinear oscillator network.
• Generative Adversarial Learning from Deterministic Processes — Generative adversarial learning can effectively model chaotic dynamical systems from deterministic time series data.
• Large-Step Training Dynamics of a Two-Factor Linear Transformer Model — Large learning rates can shift transformer training from convergence to cycles or chaos, not just accelerate it.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.