Robust DNN Surrogate Models with Uncertainty Quantification via Adversarial Training
This addresses a critical issue for researchers and practitioners using DNN surrogates in scientific simulations, where robustness is essential for reliable UQ, though it is an incremental application of existing adversarial training techniques.
The paper tackled the problem of deep neural network (DNN) surrogate models being sensitive to input perturbations, which can undermine uncertainty quantification (UQ) in simulations, and showed that adversarial training methods significantly improve robustness without losing accuracy.
For computational efficiency, surrogate models have been used to emulate mathematical simulators for physical or biological processes. High-speed simulation is crucial for conducting uncertainty quantification (UQ) when the simulation is repeated over many randomly sampled input points (aka, the Monte Carlo method). In some cases, UQ is only feasible with a surrogate model. Recently, Deep Neural Network (DNN) surrogate models have gained popularity for their hard-to-match emulation accuracy. However, it is well-known that DNN is prone to errors when input data are perturbed in particular ways, the very motivation for adversarial training. In the usage scenario of surrogate models, the concern is less of a deliberate attack but more of the high sensitivity of the DNN's accuracy to input directions, an issue largely ignored by researchers using emulation models. In this paper, we show the severity of this issue through empirical studies and hypothesis testing. Furthermore, we adopt methods in adversarial training to enhance the robustness of DNN surrogate models. Experiments demonstrate that our approaches significantly improve the robustness of the surrogate models without compromising emulation accuracy.