Beyond Uniform Sampling: Synergistic Active Learning and Input Denoising for Robust Neural Operators

ArXi:2604.13316v1 Announce Type: cross Neural operators have emerged as fast surrogate models for physics simulations, yet they remain acutely vulnerable to adversarial perturbations, a critical liability for safety-critical digital twin deployments. We present a synergistic defense that combines active learning-based data generation with an input denoising architecture. The active learning component adaptively probes model weaknesses using differential evolution attacks, then generates targeted.