Jeffreys versus Shtarkov distributions associated with some natural exponential families
نویسندگان
چکیده
Jeffreys and Shtarkov distributions play an important role in universal coding andminimumdescription length (MDL) inference, two central areas within the field of information theory. It was recently discovered that in some situations Shtarkov distributions exist while Jeffreys distributions do not. To demonstrate some of these situations we consider in this note the class of natural exponential families (NEF’s) and present a general result which enables us to construct numerous classes of infinitely divisible NEF’s for which Shtarkov distributions exist and Jeffreys distributions do not. Themethod used to obtain our general results is based on the variance functions of such NEF’s. We first present two classes of parametric NEF’s demonstrating our general results and then generalize them to obtain numerous multiparameter classes of the same type. © 2010 Elsevier B.V. All rights reserved.
منابع مشابه
Horizon-Independent Optimal Prediction with Log-Loss in Exponential Families
We study online learning under logarithmic loss with regular parametric models. Hedayati and Bartlett (2012b) showed that a Bayesian prediction strategy with Jeffreys prior and sequential normalized maximum likelihood (SNML) coincide and are optimal if and only if the latter is exchangeable, and if and only if the optimal strategy can be calculated without knowing the time horizon in advance. T...
متن کاملInvariant conjugate analysis for exponential families
There are several ways to parameterize a distribution belonging to an exponential family, each one leading to a different Bayesian analysis of the data under standard conjugate priors. To overcome this problem, we propose a new class of conjugate priors which is invariant with respect to smooth reparameterization. This class of priors contains the Jeffreys prior as a special case, according to ...
متن کاملRegret and Jeffreys Integrals in Exp. Families
where Z is the partition function Z( ) = R exp( x) dQx, and can := f j Z( ) < 1g is the canonical parameter space. We let sup = supf j 2 cang, and inf likewise. The elements of the exponential family are also parametrized by their mean value . We write for the mean value corresponding to the canonical parameter and for the canonical parameter corresponding to the mean value : For any x the maxi...
متن کاملThompson sampling for one-dimensional exponential family bandits
Thompson Sampling has been demonstrated in many complex bandit models, however the theoretical guarantees available for the parametric multi-armed bandit are still limited to the Bernoulli case. Here we extend them by proving asymptotic optimality of the algorithm using the Jeffreys prior for 1-dimensional exponential family bandits. Our proof builds on previous work, but also makes extensive u...
متن کاملTruncated Linear Minimax Estimator of a Power of the Scale Parameter in a Lower- Bounded Parameter Space
Minimax estimation problems with restricted parameter space reached increasing interest within the last two decades Some authors derived minimax and admissible estimators of bounded parameters under squared error loss and scale invariant squared error loss In some truncated estimation problems the most natural estimator to be considered is the truncated version of a classic...
متن کامل