Review for NeurIPS paper: Stationary Activations for Uncertainty Calibration in Deep Learning