How would you approach uncertainty estimation in a deep learning model for safety-critical automotive applications?
