A model-agnostic framework that enhances the performance of both model-based and model-free network inference methods when dealing with noisy measurement data.