TAGS: Uncertainty Estimation Bayesian Inference Active Learning Reinforcement Learning
ICLR 2020 brought a lot of valuable insights about uncertainty of modern deep learning models and in this post I would like to describe shortly the most interesting papers which were presented during the conference.
Recent advances in areas of Deep Learning and Neural Networks have solved a large number of different problems in a dozen of areas: computer vision, natural language processing, reinforcement learning, speech recognition and etc. The widespread availability of computational resources made it possible even for non-professionals to develop complex and sophisticated models. But do we really understand how do they work? Uncertainty Estimation is one of the ways to figure that our.