On a Problem of Robbins

نویسندگان

  • Jiaying Gu
  • Roger Koenker
چکیده

Robbins’s visionary 1951 paper can be seen as an exercise in binary classification, but also as a precursor to the outpouring of recent work on high-dimensional data analysis and multiple testing. It can also be seen as the birth of empirical Bayes methods. Our objective in the present note is to use this problem and several variants of it to provide a glimpse into the evolution of empirical Bayes methods. Much more comprehensive surveys of empirical Bayes methods and their modern relevance are provided by Zhang (2003) and Efron (2010); here, we aspire only to tell a more condensed version of the story, but one that highlights the critical role that the nonparametric maximum likelihood estimator (NPMLE) of Kiefer & Wolfowitz (1956) can play. Recent developments in convex optimization, as argued in Koenker & Mizera (2014), have greatly expanded the applicability of the Kiefer–Wolfowitz estimator and thereby increased the potential scope of nonparametric empirical Bayes methods. In prior work, Koenker & Mizera (2014), Koenker (2014), Koenker & Gu (2013), and Gu & Koenker (2014), we have emphasized the role of the Kiefer–Wolfowitz NPMLE in various estimation problems typically under squared-error loss. In this paper, in contrast, we will stress its potential usefulness mainly in classification and multiple testing.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A Continuous-time Approach to Robbins’ Problem of Minimizing the Expected Rank

Let X1, X2, . . . , Xn be independent random variables uniformly distributed on [0, 1]. We observe these sequentially and have to stop on exactly one of them. No recall of preceding observations is permitted. What stopping rule minimizes the expected rank of the selected observation? What is the value of the expected rank (as a function of n) and what is the limit of this value when n goes to ∞...

متن کامل

The Robbins Equation

Researchers have long been interested in nding simple axiomatizations of theories, including Boolean algebra. Boolean algebra can be axiomatized in many diierent ways using various sets of operations. The ordinary operations of Boolean algebra are disjunction (+), conjunction (), negation (0), zero (0), and one (1). One of the rst simple axiomatizations was presented by E. V. (Associativity) ((...

متن کامل

Investigating the relationship between safety and organizational culture in employees of an automobile company

goal of safety activities are focused on reducing the risk of the workplace, and very few activities are focused on increasing the organizational culture and safety. The purpose of this study is to examine the relationship between organizational culture and safety in one of the largest automobile companies in the country. Materials and Methods The current research was a correlational study tha...

متن کامل

The Blind Passenger and the Assignment Problem

We introduce a discrete random process which we call the passenger model, and show that it is connected to a certain random model of the assignment problem and in particular to the so-called Buck–Chan–Robbins urn process. We propose a conjecture on the distribution of the location of the minimum cost assignment in a cost matrix with zeros at specified positions and remaining entries of exponent...

متن کامل

Half-Prophets and Robbins’ Problem of Minimizing the Expected Rank

Let X1,X2, . . . ,Xn be i.i.d. random variables with a known continuous distribution function. Robbins’ problem is to find a sequential stopping rule without recall which minimizes the expected rank of the selected observation. An upper bound (obtained by memoryless threshold rules) and a procedure to obtain lower bounds of the value are known, but the essence of the problem is still unsolved. ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2015