Statistics Seminars: Introduction to conformal prediction
2 February 2015 02:05 in CM221
Conformal prediction is a method, closely related to nonparametric predictive inference, of producing prediction sets that can be applied on top of a wide range of prediction algorithms. The method has a guaranteed coverage probability under the standard IID assumption regardless of whether the assumptions (often considerably more restrictive) of the underlying algorithm are satisfied. In this talk I will review and compare conformal prediction and nonparametric predictive inference. If time allows, I will also state and discuss several recent results about conformal prediction that at this time do not have counterparts in nonparametric predictive inference. An example of such a result is that the conformal predictor based on Bayesian ridge regression loses little in efficiency as compared with the underlying algorithm when the latter's assumptions are satisfied (whereas being a conformal predictor, it has the stronger guarantee of validity).
Contact firstname.lastname@example.org for more information