Technology Category
Search and Optimization: Learning to SearchMachine Learning: Calibration & Uncertainty QuantificationReasoning under Uncertainty: Relational Probabilistic Models
A data augmentation strategy for deep neural networks with application to epidemic modelling
🤖 AI Summary
To address the high computational cost and poor scalability of physics-informed neural networks (PINNs) in epidemic forecasting, this paper proposes a novel paradigm integrating epidemiological dynamics priors with data-driven modeling. Methodologically, we first embed constraints derived from SIR-type models—including a saturated infection rate mechanism—into a data augmentation strategy, thereby systematically enhancing the reliability and generalization capability of feedforward neural networks (FNNs) and nonlinear autoregressive (NAR) models for nonlinear epidemic dynamics. Our key contribution is a lightweight, purely data-driven model that approximates PINN-level accuracy without solving partial differential equations or backpropagating physical residuals. Experiments on real-world post-lockdown COVID-19 data from Italy and Spain demonstrate significant improvements in short-term prediction accuracy and robustness, achieving a favorable trade-off among high precision, low computational overhead, and strong scalability—offering an efficient and practical alternative for data-driven epidemiological modeling.
Technology Category
Application Category
📝 Abstract
In this work, we integrate the predictive capabilities of compartmental disease dynamics models with machine learning ability to analyze complex, high-dimensional data and uncover patterns that conventional models may overlook. Specifically, we present a proof of concept demonstrating the application of data-driven methods and deep neural networks to a recently introduced SIR-type model with social features, including a saturated incidence rate, to improve epidemic prediction and forecasting. Our results show that a robust data augmentation strategy trough suitable data-driven models can improve the reliability of Feed-Forward Neural Networks (FNNs) and Nonlinear Autoregressive Networks (NARs), making them viable alternatives to Physics-Informed Neural Networks (PINNs). This approach enhances the ability to handle nonlinear dynamics and offers scalable, data-driven solutions for epidemic forecasting, prioritizing predictive accuracy over the constraints of physics-based models. Numerical simulations of the post-lockdown phase of the COVID-19 epidemic in Italy and Spain validate our methodology.
Problem
Research questions and friction points this paper is trying to address.
Enhance epidemic prediction using deep neural networks.
Integrate compartmental disease models with machine learning.
Improve forecasting reliability with data augmentation strategies.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Integrates compartmental models with deep learning
Uses data augmentation for epidemic prediction
Enhances FNNs and NARs reliability over PINNs
🔎 Similar Papers
No similar papers found.
