Rare Probability Estimation under Regularly Varying Heavy Tails

Mesrob I. Ohannessian, Munther A. Dahleh
Proceedings of the 25th Annual Conference on Learning Theory, PMLR 23:21.1-21.24, 2012.

Abstract

This paper studies the problem of estimating the probability of symbols that have occurred very rarely, in samples drawn independently from an unknown, possibly infinite, discrete distribution. In particular, we study the multiplicative consistency of estimators, defined as the ratio of the estimate to the true quantity converging to one. We first show that the classical Good-Turing estimator is not universally consistent in this sense, despite enjoying favorable additive properties. We then use Karamata’s theory of regular variation to prove that regularly varying heavy tails are sufficient for consistency. At the core of this result is a multiplicative concentration that we establish both by extending the McAllester-Ortiz additive concentration for the missing mass to all rare probabilities and by exploiting regular variation. We also derive a family of estimators which, in addition to being consistent, address some of the shortcomings of the Good-Turing estimator. For example, they perform smoothing implicitly and have the absolute discounting structure of many heuristic algorithms. This also establishes a discrete parallel to extreme value theory, and many of the techniques therein can be adapted to the framework that we set forth.

Cite this Paper


BibTeX
@InProceedings{pmlr-v23-ohannessian12, title = {Rare Probability Estimation under Regularly Varying Heavy Tails}, author = {Ohannessian, Mesrob I. and Dahleh, Munther A.}, booktitle = {Proceedings of the 25th Annual Conference on Learning Theory}, pages = {21.1--21.24}, year = {2012}, editor = {Mannor, Shie and Srebro, Nathan and Williamson, Robert C.}, volume = {23}, series = {Proceedings of Machine Learning Research}, address = {Edinburgh, Scotland}, month = {25--27 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v23/ohannessian12/ohannessian12.pdf}, url = {https://proceedings.mlr.press/v23/ohannessian12.html}, abstract = {This paper studies the problem of estimating the probability of symbols that have occurred very rarely, in samples drawn independently from an unknown, possibly infinite, discrete distribution. In particular, we study the multiplicative consistency of estimators, defined as the ratio of the estimate to the true quantity converging to one. We first show that the classical Good-Turing estimator is not universally consistent in this sense, despite enjoying favorable additive properties. We then use Karamata’s theory of regular variation to prove that regularly varying heavy tails are sufficient for consistency. At the core of this result is a multiplicative concentration that we establish both by extending the McAllester-Ortiz additive concentration for the missing mass to all rare probabilities and by exploiting regular variation. We also derive a family of estimators which, in addition to being consistent, address some of the shortcomings of the Good-Turing estimator. For example, they perform smoothing implicitly and have the absolute discounting structure of many heuristic algorithms. This also establishes a discrete parallel to extreme value theory, and many of the techniques therein can be adapted to the framework that we set forth.} }
Endnote
%0 Conference Paper %T Rare Probability Estimation under Regularly Varying Heavy Tails %A Mesrob I. Ohannessian %A Munther A. Dahleh %B Proceedings of the 25th Annual Conference on Learning Theory %C Proceedings of Machine Learning Research %D 2012 %E Shie Mannor %E Nathan Srebro %E Robert C. Williamson %F pmlr-v23-ohannessian12 %I PMLR %P 21.1--21.24 %U https://proceedings.mlr.press/v23/ohannessian12.html %V 23 %X This paper studies the problem of estimating the probability of symbols that have occurred very rarely, in samples drawn independently from an unknown, possibly infinite, discrete distribution. In particular, we study the multiplicative consistency of estimators, defined as the ratio of the estimate to the true quantity converging to one. We first show that the classical Good-Turing estimator is not universally consistent in this sense, despite enjoying favorable additive properties. We then use Karamata’s theory of regular variation to prove that regularly varying heavy tails are sufficient for consistency. At the core of this result is a multiplicative concentration that we establish both by extending the McAllester-Ortiz additive concentration for the missing mass to all rare probabilities and by exploiting regular variation. We also derive a family of estimators which, in addition to being consistent, address some of the shortcomings of the Good-Turing estimator. For example, they perform smoothing implicitly and have the absolute discounting structure of many heuristic algorithms. This also establishes a discrete parallel to extreme value theory, and many of the techniques therein can be adapted to the framework that we set forth.
RIS
TY - CPAPER TI - Rare Probability Estimation under Regularly Varying Heavy Tails AU - Mesrob I. Ohannessian AU - Munther A. Dahleh BT - Proceedings of the 25th Annual Conference on Learning Theory DA - 2012/06/16 ED - Shie Mannor ED - Nathan Srebro ED - Robert C. Williamson ID - pmlr-v23-ohannessian12 PB - PMLR DP - Proceedings of Machine Learning Research VL - 23 SP - 21.1 EP - 21.24 L1 - http://proceedings.mlr.press/v23/ohannessian12/ohannessian12.pdf UR - https://proceedings.mlr.press/v23/ohannessian12.html AB - This paper studies the problem of estimating the probability of symbols that have occurred very rarely, in samples drawn independently from an unknown, possibly infinite, discrete distribution. In particular, we study the multiplicative consistency of estimators, defined as the ratio of the estimate to the true quantity converging to one. We first show that the classical Good-Turing estimator is not universally consistent in this sense, despite enjoying favorable additive properties. We then use Karamata’s theory of regular variation to prove that regularly varying heavy tails are sufficient for consistency. At the core of this result is a multiplicative concentration that we establish both by extending the McAllester-Ortiz additive concentration for the missing mass to all rare probabilities and by exploiting regular variation. We also derive a family of estimators which, in addition to being consistent, address some of the shortcomings of the Good-Turing estimator. For example, they perform smoothing implicitly and have the absolute discounting structure of many heuristic algorithms. This also establishes a discrete parallel to extreme value theory, and many of the techniques therein can be adapted to the framework that we set forth. ER -
APA
Ohannessian, M.I. & Dahleh, M.A.. (2012). Rare Probability Estimation under Regularly Varying Heavy Tails. Proceedings of the 25th Annual Conference on Learning Theory, in Proceedings of Machine Learning Research 23:21.1-21.24 Available from https://proceedings.mlr.press/v23/ohannessian12.html.

Related Material