Towards calibrated and scalable uncertainty representations for neural networks

Nabeel Seedat, Christopher Kanan

For many applications it is critical to know the uncertainty of a neural network's predictions. While a variety of neural network parameter estimation methods have been proposed fo…