Hyperparameter tuning improves equation learning in biological neural networks
Hyperparameter selection for equation learning with biologically-informed neural networks
Machine Learning
Summary
Learning the hidden rules behind complex biological systems can be very hard, especially when data is messy or incomplete. The authors studied a type of neural network designed to find these rules from data, called biologically-informed neural networks (BINNs). They focused on how to choose important settings within these networks, called hyperparameters, that control learning quality. Their study provides a step-by-step process and practical advice for selecting these settings, helping users get better results even when they don’t know the true underlying biological equations.
What this means in practice
- •For biomedical modelers: Improve model fitting for biological systems with sparse and noisy data using structured hyperparameter tuning.
- •For computational biologists: Use guided hyperparameter selection workflow to enhance discovery of unknown biological equations from experimental data.
Authors
William Lavery, Jodie A. Cochrane, John T. Nardini, Sara Hamis
Abstract
Biologically-informed neural networks (BINNs) have emerged as a flexible subclass of physics-informed neural networks (PINNs) for learning terms in partial differential equations from data. BINNs are particularly suited for biological systems, where the governing equations are highly nonlinear and only partially known a priori, and where data observations are often sparse, noisy, and incomplete. However, applying BINNs effectively in practice depends critically on hyperparameter selection, which remains a central challenge in equation-learning frameworks. Hyperparameters are often chosen heuristically and only cursorily documented, which limits the reproducibility of results and the transferability of methods. We present a diagnostic workflow for hyperparameter selection that can be used when the ground-truth equations are not known. The workflow is guided by three main questions: (1) Are the benefits of greater network capacity worth the cost? (2) Do more training epochs keep reducing the validation loss? (3) Do the learned terms stop changing as network capacity and training increase? We apply our workflow to synthetic systems of varying complexity with known ground truth, spanning diffusion and growth right-hand side terms and data ranging from 1D+t to 2D+t. We demonstrate that the validation loss generally follows the true error in the learned terms and distil practical rules of thumb for selecting hyperparameters in the BINN architecture. By providing a structured workflow, practical guidelines, and suggested starting values for hyperparameter selection, this work lowers the barrier to BINN-based equation learning.