Classical classification models allow to estimate the quality of predictions by associating their output with probability values. However, they do not allow to construct prediction models that are guaranteed to produce output with a specific level of accuracy. In this talk, we will discuss a framework of Conformal learning — an approach that produces prediction with guarantees.
We will talk about the main ideas and trends in Conformal learning, and the price that we have to pay for predictions with guarantees — a reduction in precision.
The webinar is free, registration at the link: