English

Agnostic insurability of model classes

Statistics Theory 2013-05-02 v3 Information Theory math.IT Statistics Theory

Abstract

Motivated by problems in insurance, our task is to predict finite upper bounds on a future draw from an unknown distribution pp over the set of natural numbers. We can only use past observations generated independently and identically distributed according to pp. While pp is unknown, it is known to belong to a given collection P{\cal P} of probability distributions on the natural numbers. The support of the distributions pPp \in {\cal P} may be unbounded, and the prediction game goes on for \emph{infinitely} many draws. We are allowed to make observations without predicting upper bounds for some time. But we must, with probability 1, start and then continue to predict upper bounds after a finite time irrespective of which pPp \in {\cal P} governs the data. If it is possible, without knowledge of pp and for any prescribed confidence however close to 1, to come up with a sequence of upper bounds that is never violated over an infinite time window with confidence at least as big as prescribed, we say the model class P{\cal P} is \emph{insurable}. We completely characterize the insurability of any class P{\cal P} of distributions over natural numbers by means of a condition on how the neighborhoods of distributions in P{\cal P} should be, one that is both necessary and sufficient.

Keywords

Cite

@article{arxiv.1212.3866,
  title  = {Agnostic insurability of model classes},
  author = {Narayana Santhanam and Venkat Anantharam},
  journal= {arXiv preprint arXiv:1212.3866},
  year   = {2013}
}
R2 v1 2026-06-21T22:55:22.060Z