RISR: Residual-Informed Scientific Equation Discovery with Large Language Models

Key points

• Symbolic regression combines structural search with numerical fitting, but aggregate fit scores do not describe how the remaining error varies across inputs.1
• A residual encoder compresses aligned inputs, targets, current predictions, and residuals into continuous tokens that condition a language model to propose formulas.1
• For subsequent refinement, a dual-view relational encoder uses additive and regularized multiplicative residuals to predict the post-fit utility of candidate corrections.1
• The results show that our residual-informed approach can improve numerical equation recovery.1

Sources (1)

• [1]RISR: Residual-Informed Scientific Equation Discovery with Large Language Models
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 07:17 AM

  We introduce RISR, a residual-informed method that uses these error patterns to guide formula discovery and learn which corrections are worth fitting.

  Symbolic regression combines structural search with numerical fitting, but aggregate fit scores do not describe how the remaining error varies across inputs.

Extractive summary: sentences quoted from the sources.