Estimating and Controlling the Uncertainty of Learning Machines

The problem of estimating model uncertainty of learning machines (LMs) is becoming a subject of great interest because of the wide application of such kind of methodologies for solving real-world problems. In this work we will provide a general overview on estimating and controlling uncertainity of LMs, by describing the algorithms, the theory and the empirical methods used to obtain a robust estimation. In the end we address the problem of uncertainty estimation when devices with limited resources are considered for the hardware implementation.