New neural network infers hidden physics from position data

Researchers have developed History-informed Lagrangian Neural Networks (HiLNN) to improve the long-term forecasting of mechanical systems using only position data. This new model addresses limitations of existing physics-guided networks by inferring hidden velocities and system parameters from historical observations. HiLNN utilizes a recurrent encoder to extract latent context, which then adapts the system's mass matrix, potential energy, and damping coefficients. The model is trained end-to-end using a differentiable Runge-Kutta method and has demonstrated superior accuracy and energy profile maintenance compared to current state-of-the-art methods. AI

IMPACT This research could lead to more accurate and adaptable forecasting models for complex mechanical systems, potentially impacting fields like robotics and autonomous systems.

RANK_REASON The cluster contains an academic paper detailing a new model architecture and its evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• classic Runge–Kutta method
• HiLNN
• History-informed Lagrangian Neural Networks
• Lagrangian neural networks

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →