|
In probability theory and statistics, the Dirichlet process (DP) is one of the most popular Bayesian nonparametric models. It was introduced by Ferguson as a prior over probability distributions. A Dirichlet process is completely defined by its parameters: (the ''base distribution'' or ''base measure'') is an arbitrary distribution and (the ''concentration parameter'') is a positive real number (it is often denoted as ). According to the Bayesian paradigma these parameters should be chosen based on the available prior information on the domain. The question is: how should we choose the prior parameters of the DP, in particular the infinite dimensional one , in case of lack of prior information? To address this issue, the only prior that has been proposed so far is the limiting DP obtained for , which has been introduced under the name of Bayesian bootstrap by Rubin;〔Rubin D (1981). The Bayesian bootstrap. Ann Statist 9 130–134〕 in fact it can be proven that the Bayesian bootstrap is asymptotically equivalent to the frequentist bootstrap introduced by Bradley Efron.〔Efron B (1979). Bootstrap methods: Another look at the jackknife. Ann. Statist. 7 1–26〕 The limiting Dirichlet process has been criticized on diverse grounds. From an a-priori point of view, the main criticism is that taking is far from leading to a noninformative prior. Moreover, a-posteriori, it assigns zero probability to any set that does not include the observations.〔Rubin D (1981). The Bayesian bootstrap. Ann Statist 9 130–134〕 The imprecise Dirichlet process has been proposed to overcome these issues. The basic idea is to fix but do not choose any precise base measure . More precisely, the imprecise Dirichlet process (IDP) is defined as follows: : where is the set of all probability measures. In other words, the IDP is the set of all Dirichlet processes (with a fixed ) obtained by letting the base measure to span the set of all probability measures. == Inferences with the Imprecise Dirichlet Process == Let a probability distribution on (here is a standard Borel space with Borel -field ) and assume that . Then consider a real-valued bounded function defined on . It is well known that the expectation of with respect to the Dirichlet process is : One of the most remarkable properties of the DP priors is that the posterior distribution of is again a DP. Let be an independent and identically distributed sample from and , then the posterior distribution of given the observations is : where is an atomic probability measure (Dirac's delta) centered at . Hence, it follows that Therefore, for any fixed , we can exploit the previous equations to derive prior and posterior expectations. In the IDP can span the set of all distributions . This implies that we will get a different prior and posterior expectation of for any choice of . A way to characterize inferences for the IDP is by computing lower and upper bounds for the expectation of w.r.t. . A-priori these bounds are: : the lower (upper) bound is obtained by a probability measure that puts all the mass on the infimum (supremum) of , i.e., with (or respectively with ). From the above expressions of the lower and upper bounds, it can be observed that the range of under the IDP is the same as the original range of . In other words, by specifying the IDP, we are not giving any prior information on the value of the expectation of . A-priori, IDP is therefore a model of prior (near)-ignorance for . A-posteriori, IDP can learn from data. The posterior lower and upper bounds for the expectation of are in fact given by: : It can be observed that the posterior inferences do not depend on . To define the IDP, the modeler has only to choose (the concentration parameter). This explains the meaning of the adjective ''near'' in prior near-ignorance, because the IDP requires by the modeller the elicitation of a parameter. However, this is a simple elicitation problem for a nonparametric prior, since we only have to choose the value of a positive scalar (there are not infinitely many parameters left in the IDP model). Finally, observe that for , IDP satisfies : where . In other words, the IDP is consistent. 抄文引用元・出典: フリー百科事典『 ウィキペディア(Wikipedia)』 ■ウィキペディアで「Imprecise Dirichlet process」の詳細全文を読む スポンサード リンク
|