Trajectory-level self-supervision for simulation driven estimators

Recent advancements in modeling have led to the construction of high-fidelity simulators (digital twins) to represent physical systems. However, the parameters of these high fidelity-simulators must be calibrated to match a given physical system. This motivated the construction of simulation-driven parameter estimators, built by generating synthetic observations for sampled parameter values and learning a supervised mapping from observations to parameters. However, when the parameters of the physical system lie outside the sampled range, predictions suffer from an out-of-distribution (OOD) error. This paper introduces a fine-tuning approach based on trajectory-level self-supervision for the Two-Stage approach, a simulation-driven estimator, that mitigates OOD effects and improves its accuracy. The effectiveness of the proposed method is verified through numerical simulations.