Andrey Malinin: Motivation and Sources of Uncertainty

Data Fest Online 2020 Uncertainty Estimation in ML track Speaker: Andrey Malinin, Yandex & HSE In this video we will motivate the need for uncertainty estimation using a number of real-world examples and discuss the two sources of uncertainty - data (aleatoric) uncertainty and knowledge (epistemic) uncertainty. Data uncertainty arises due to input in regions of class overlap and additive noise, while knowledge uncertainty arises due to input coming from regions which are either sparsely covered by training data, or not covered by training data at all. Register and get access to the tracks: Join the community: