f -Cal: Calibrated aleatoric uncertainty estimation from neural networks for robot perception

fcal

Abstract

f-Cal is calibration method for probabilistic regression networks. Typical Bayesian neural networks are overconfident in their predictions. For these predictions to be used in downstream tasks, reliable and calibrated uncertainity estimates are critical. f-Cal proposes a simple loss function to remedy this; this can be employed to train any probabilistic neural regressor to produced calibrated estimates of aleatoric uncertainty.

Publication
In International Conference on Robotics and Automation
Dhaivat Bhatt
Masters Student
Krishna Murthy Jatavallabhula
Krishna Murthy Jatavallabhula
Postdoc

My research builds “world models” – a necessity for intelligent embodied agents acting in the real world. My work spans robotics, computer vision, graphics, and deep learning.

Liam Paull
Assistant Professor

I lead the Montreal robotics and embodied AI lab. I am affiliated with Université de Montréal, Mila, and I hold a CIFAR AI chair.

comments powered by Disqus

Related