Learning epidemic trajectories through Kernel Operator Learning: from modelling to optimal control

• DOI: 10.48550/arxiv.2404.11130
• Publication Date: 2024-04-17
• Authors: Giovanni Ziarelli; Nicola Parolini; Marco Verani
• Keywords: Operator (biology); Kernel (algebra); Multiple kernel learning; Epidemic control

Since infectious pathogens start spreading into a susceptible population, mathematical models can provide policy makers with reliable forecasts and scenario analyses, which be concretely implemented or solely consulted. In these complex epidemiological scenarios, machine learning architectures play an important role, since they directly reconstruct data-driven circumventing the specific modelling choices parameter calibration, typical of classical compartmental models. this work, we discuss efficacy Kernel Operator Learning (KOL) to population dynamics during epidemic outbreaks, where transmission rate is ruled by input strategy. particular, introduce two surrogate models, named KOL-m KOL-$\partial$, in different ways evolution epidemics. Moreover, evaluate generalization performances approaches kernels, including Neural Tangent Kernels, compare them neural network model method. Employing synthetic but semi-realistic data, show how introduced are suitable for realizing fast robust competitive determining optimal intervention strategies respect performance measures.