andrew_gelman_stats andrew_gelman_stats-2012 andrew_gelman_stats-2012-1454 knowledge-graph by maker-knowledge-mining
Source: html
Introduction: Nathaniel Egwu writes: I am a PhD student working on machine learning using artificial neural networks . . . Do you have some recent publications related to how one can construct priors depending on the type of input data available for training? I intend to construct a prior distribution for a given trade-off parameter of my non model obtained through training a neural network. At this stage, my argument is due to the fact that Bayesian nonparameteric estimation offers some insight on how to proceed on this problem. As I’ve been writing here for awhile, I’ve been interested in weakly informative priors. But I have little experience with nonparametric models. Perhaps Aki Vehtari or David Dunson or some other expert on these models can discuss how to set them up with weakly informative priors? This sounds like it could be important to me.
sentIndex sentText sentNum sentScore
1 Nathaniel Egwu writes: I am a PhD student working on machine learning using artificial neural networks . [sent-1, score-0.995]
2 Do you have some recent publications related to how one can construct priors depending on the type of input data available for training? [sent-4, score-1.202]
3 I intend to construct a prior distribution for a given trade-off parameter of my non model obtained through training a neural network. [sent-5, score-1.578]
4 At this stage, my argument is due to the fact that Bayesian nonparameteric estimation offers some insight on how to proceed on this problem. [sent-6, score-0.783]
5 As I’ve been writing here for awhile, I’ve been interested in weakly informative priors. [sent-7, score-0.604]
6 But I have little experience with nonparametric models. [sent-8, score-0.29]
7 Perhaps Aki Vehtari or David Dunson or some other expert on these models can discuss how to set them up with weakly informative priors? [sent-9, score-0.722]
wordName wordTfidf (topN-words)
[('neural', 0.345), ('construct', 0.279), ('weakly', 0.273), ('training', 0.243), ('priors', 0.205), ('informative', 0.2), ('vehtari', 0.191), ('dunson', 0.18), ('intend', 0.166), ('non', 0.164), ('artificial', 0.157), ('proceed', 0.155), ('aki', 0.153), ('nonparametric', 0.14), ('offers', 0.14), ('depending', 0.138), ('stage', 0.137), ('obtained', 0.136), ('phd', 0.135), ('input', 0.134), ('insight', 0.13), ('networks', 0.127), ('publications', 0.127), ('machine', 0.118), ('awhile', 0.106), ('expert', 0.104), ('estimation', 0.103), ('due', 0.102), ('sounds', 0.098), ('parameter', 0.095), ('type', 0.093), ('learning', 0.093), ('student', 0.089), ('available', 0.088), ('experience', 0.086), ('discuss', 0.083), ('related', 0.082), ('argument', 0.082), ('david', 0.08), ('distribution', 0.076), ('prior', 0.074), ('ve', 0.074), ('fact', 0.071), ('writing', 0.066), ('working', 0.066), ('interested', 0.065), ('little', 0.064), ('set', 0.062), ('bayesian', 0.057), ('recent', 0.056)]
simIndex simValue blogId blogTitle
same-blog 1 1.0 1454 andrew gelman stats-2012-08-11-Weakly informative priors for Bayesian nonparametric models?
Introduction: Nathaniel Egwu writes: I am a PhD student working on machine learning using artificial neural networks . . . Do you have some recent publications related to how one can construct priors depending on the type of input data available for training? I intend to construct a prior distribution for a given trade-off parameter of my non model obtained through training a neural network. At this stage, my argument is due to the fact that Bayesian nonparameteric estimation offers some insight on how to proceed on this problem. As I’ve been writing here for awhile, I’ve been interested in weakly informative priors. But I have little experience with nonparametric models. Perhaps Aki Vehtari or David Dunson or some other expert on these models can discuss how to set them up with weakly informative priors? This sounds like it could be important to me.
Introduction: Deborah Mayo sent me this quote from Jim Berger: Too often I see people pretending to be subjectivists, and then using “weakly informative” priors that the objective Bayesian community knows are terrible and will give ridiculous answers; subjectivism is then being used as a shield to hide ignorance. . . . In my own more provocative moments, I claim that the only true subjectivists are the objective Bayesians, because they refuse to use subjectivism as a shield against criticism of sloppy pseudo-Bayesian practice. This caught my attention because I’ve become more and more convinced that weakly informative priors are the right way to go in many different situations. I don’t think Berger was talking about me , though, as the above quote came from a publication in 2006, at which time I’d only started writing about weakly informative priors. Going back to Berger’s article , I see that his “weakly informative priors” remark was aimed at this article by Anthony O’Hagan, who w
3 0.28803426 1092 andrew gelman stats-2011-12-29-More by Berger and me on weakly informative priors
Introduction: A couple days ago we discussed some remarks by Tony O’Hagan and Jim Berger on weakly informative priors. Jim followed up on Deborah Mayo’s blog with this: Objective Bayesian priors are often improper (i.e., have infinite total mass), but this is not a problem when they are developed correctly. But not every improper prior is satisfactory. For instance, the constant prior is known to be unsatisfactory in many situations. The ‘solution’ pseudo-Bayesians often use is to choose a constant prior over a large but bounded set (a ‘weakly informative’ prior), saying it is now proper and so all is well. This is not true; if the constant prior on the whole parameter space is bad, so will be the constant prior over the bounded set. The problem is, in part, that some people confuse proper priors with subjective priors and, having learned that true subjective priors are fine, incorrectly presume that weakly informative proper priors are fine. I have a few reactions to this: 1. I agree
4 0.26485059 1858 andrew gelman stats-2013-05-15-Reputations changeable, situations tolerable
Introduction: David Kessler, Peter Hoff, and David Dunson write : Marginally specified priors for nonparametric Bayesian estimation Prior specification for nonparametric Bayesian inference involves the difficult task of quantifying prior knowledge about a parameter of high, often infinite, dimension. Realistically, a statistician is unlikely to have informed opinions about all aspects of such a parameter, but may have real information about functionals of the parameter, such the population mean or variance. This article proposes a new framework for nonparametric Bayes inference in which the prior distribution for a possibly infinite-dimensional parameter is decomposed into two parts: an informative prior on a finite set of functionals, and a nonparametric conditional prior for the parameter given the functionals. Such priors can be easily constructed from standard nonparametric prior distributions in common use, and inherit the large support of the standard priors upon which they are based. Ad
5 0.2011044 846 andrew gelman stats-2011-08-09-Default priors update?
Introduction: Ryan King writes: I was wondering if you have a brief comment on the state of the art for objective priors for hierarchical generalized linear models (generalized linear mixed models). I have been working off the papers in Bayesian Analysis (2006) 1, Number 3 (Browne and Draper, Kass and Natarajan, Gelman). There seems to have been continuous work for matching priors in linear mixed models, but GLMMs less so because of the lack of an analytic marginal likelihood for the variance components. There are a number of additional suggestions in the literature since 2006, but little robust practical guidance. I’m interested in both mean parameters and the variance components. I’m almost always concerned with logistic random effect models. I’m fascinated by the matching-priors idea of higher-order asymptotic improvements to maximum likelihood, and need to make some kind of defensible default recommendation. Given the massive scale of the datasets (genetics …), extensive sensitivity a
6 0.15674463 903 andrew gelman stats-2011-09-13-Duke postdoctoral fellowships in nonparametric Bayes & high-dimensional data
7 0.15640891 1486 andrew gelman stats-2012-09-07-Prior distributions for regression coefficients
8 0.1347907 1209 andrew gelman stats-2012-03-12-As a Bayesian I want scientists to report their data non-Bayesianly
9 0.13349625 801 andrew gelman stats-2011-07-13-On the half-Cauchy prior for a global scale parameter
10 0.13187326 2129 andrew gelman stats-2013-12-10-Cross-validation and Bayesian estimation of tuning parameters
11 0.12941939 1155 andrew gelman stats-2012-02-05-What is a prior distribution?
12 0.12748685 779 andrew gelman stats-2011-06-25-Avoiding boundary estimates using a prior distribution as regularization
14 0.12498356 468 andrew gelman stats-2010-12-15-Weakly informative priors and imprecise probabilities
15 0.12359884 1946 andrew gelman stats-2013-07-19-Prior distributions on derived quantities rather than on parameters themselves
16 0.11826608 1695 andrew gelman stats-2013-01-28-Economists argue about Bayes
17 0.11660071 1674 andrew gelman stats-2013-01-15-Prior Selection for Vector Autoregressions
18 0.11534998 2109 andrew gelman stats-2013-11-21-Hidden dangers of noninformative priors
19 0.11506166 1046 andrew gelman stats-2011-12-07-Neutral noninformative and informative conjugate beta and gamma prior distributions
20 0.1104693 1482 andrew gelman stats-2012-09-04-Model checking and model understanding in machine learning
topicId topicWeight
[(0, 0.133), (1, 0.157), (2, -0.038), (3, 0.071), (4, -0.027), (5, 0.007), (6, 0.079), (7, 0.013), (8, -0.162), (9, 0.099), (10, 0.026), (11, 0.023), (12, 0.045), (13, 0.025), (14, -0.036), (15, -0.025), (16, 0.022), (17, -0.025), (18, 0.005), (19, 0.007), (20, -0.03), (21, -0.024), (22, -0.002), (23, 0.018), (24, -0.02), (25, -0.012), (26, 0.03), (27, -0.028), (28, -0.009), (29, 0.016), (30, -0.007), (31, -0.058), (32, -0.006), (33, 0.005), (34, 0.011), (35, 0.026), (36, 0.002), (37, -0.012), (38, 0.04), (39, -0.011), (40, -0.029), (41, -0.008), (42, 0.037), (43, 0.053), (44, -0.001), (45, 0.045), (46, -0.02), (47, -0.002), (48, -0.003), (49, -0.0)]
simIndex simValue blogId blogTitle
same-blog 1 0.96845919 1454 andrew gelman stats-2012-08-11-Weakly informative priors for Bayesian nonparametric models?
Introduction: Nathaniel Egwu writes: I am a PhD student working on machine learning using artificial neural networks . . . Do you have some recent publications related to how one can construct priors depending on the type of input data available for training? I intend to construct a prior distribution for a given trade-off parameter of my non model obtained through training a neural network. At this stage, my argument is due to the fact that Bayesian nonparameteric estimation offers some insight on how to proceed on this problem. As I’ve been writing here for awhile, I’ve been interested in weakly informative priors. But I have little experience with nonparametric models. Perhaps Aki Vehtari or David Dunson or some other expert on these models can discuss how to set them up with weakly informative priors? This sounds like it could be important to me.
2 0.87822568 1858 andrew gelman stats-2013-05-15-Reputations changeable, situations tolerable
Introduction: David Kessler, Peter Hoff, and David Dunson write : Marginally specified priors for nonparametric Bayesian estimation Prior specification for nonparametric Bayesian inference involves the difficult task of quantifying prior knowledge about a parameter of high, often infinite, dimension. Realistically, a statistician is unlikely to have informed opinions about all aspects of such a parameter, but may have real information about functionals of the parameter, such the population mean or variance. This article proposes a new framework for nonparametric Bayes inference in which the prior distribution for a possibly infinite-dimensional parameter is decomposed into two parts: an informative prior on a finite set of functionals, and a nonparametric conditional prior for the parameter given the functionals. Such priors can be easily constructed from standard nonparametric prior distributions in common use, and inherit the large support of the standard priors upon which they are based. Ad
3 0.877707 1092 andrew gelman stats-2011-12-29-More by Berger and me on weakly informative priors
Introduction: A couple days ago we discussed some remarks by Tony O’Hagan and Jim Berger on weakly informative priors. Jim followed up on Deborah Mayo’s blog with this: Objective Bayesian priors are often improper (i.e., have infinite total mass), but this is not a problem when they are developed correctly. But not every improper prior is satisfactory. For instance, the constant prior is known to be unsatisfactory in many situations. The ‘solution’ pseudo-Bayesians often use is to choose a constant prior over a large but bounded set (a ‘weakly informative’ prior), saying it is now proper and so all is well. This is not true; if the constant prior on the whole parameter space is bad, so will be the constant prior over the bounded set. The problem is, in part, that some people confuse proper priors with subjective priors and, having learned that true subjective priors are fine, incorrectly presume that weakly informative proper priors are fine. I have a few reactions to this: 1. I agree
Introduction: Deborah Mayo sent me this quote from Jim Berger: Too often I see people pretending to be subjectivists, and then using “weakly informative” priors that the objective Bayesian community knows are terrible and will give ridiculous answers; subjectivism is then being used as a shield to hide ignorance. . . . In my own more provocative moments, I claim that the only true subjectivists are the objective Bayesians, because they refuse to use subjectivism as a shield against criticism of sloppy pseudo-Bayesian practice. This caught my attention because I’ve become more and more convinced that weakly informative priors are the right way to go in many different situations. I don’t think Berger was talking about me , though, as the above quote came from a publication in 2006, at which time I’d only started writing about weakly informative priors. Going back to Berger’s article , I see that his “weakly informative priors” remark was aimed at this article by Anthony O’Hagan, who w
5 0.82722658 846 andrew gelman stats-2011-08-09-Default priors update?
Introduction: Ryan King writes: I was wondering if you have a brief comment on the state of the art for objective priors for hierarchical generalized linear models (generalized linear mixed models). I have been working off the papers in Bayesian Analysis (2006) 1, Number 3 (Browne and Draper, Kass and Natarajan, Gelman). There seems to have been continuous work for matching priors in linear mixed models, but GLMMs less so because of the lack of an analytic marginal likelihood for the variance components. There are a number of additional suggestions in the literature since 2006, but little robust practical guidance. I’m interested in both mean parameters and the variance components. I’m almost always concerned with logistic random effect models. I’m fascinated by the matching-priors idea of higher-order asymptotic improvements to maximum likelihood, and need to make some kind of defensible default recommendation. Given the massive scale of the datasets (genetics …), extensive sensitivity a
6 0.82529652 468 andrew gelman stats-2010-12-15-Weakly informative priors and imprecise probabilities
7 0.81699646 1946 andrew gelman stats-2013-07-19-Prior distributions on derived quantities rather than on parameters themselves
8 0.80501968 1474 andrew gelman stats-2012-08-29-More on scaled-inverse Wishart and prior independence
9 0.80450356 801 andrew gelman stats-2011-07-13-On the half-Cauchy prior for a global scale parameter
11 0.79770118 2017 andrew gelman stats-2013-09-11-“Informative g-Priors for Logistic Regression”
12 0.79655528 1465 andrew gelman stats-2012-08-21-D. Buggin
13 0.78057539 2109 andrew gelman stats-2013-11-21-Hidden dangers of noninformative priors
14 0.77806693 1155 andrew gelman stats-2012-02-05-What is a prior distribution?
15 0.7718119 669 andrew gelman stats-2011-04-19-The mysterious Gamma (1.4, 0.4)
16 0.76469195 1486 andrew gelman stats-2012-09-07-Prior distributions for regression coefficients
17 0.75848383 779 andrew gelman stats-2011-06-25-Avoiding boundary estimates using a prior distribution as regularization
18 0.75476968 2129 andrew gelman stats-2013-12-10-Cross-validation and Bayesian estimation of tuning parameters
19 0.7275762 2138 andrew gelman stats-2013-12-18-In Memoriam Dennis Lindley
20 0.72549039 184 andrew gelman stats-2010-08-04-That half-Cauchy prior
topicId topicWeight
[(13, 0.026), (15, 0.018), (16, 0.073), (24, 0.204), (53, 0.032), (59, 0.041), (65, 0.137), (71, 0.025), (86, 0.074), (99, 0.255)]
simIndex simValue blogId blogTitle
same-blog 1 0.96383417 1454 andrew gelman stats-2012-08-11-Weakly informative priors for Bayesian nonparametric models?
Introduction: Nathaniel Egwu writes: I am a PhD student working on machine learning using artificial neural networks . . . Do you have some recent publications related to how one can construct priors depending on the type of input data available for training? I intend to construct a prior distribution for a given trade-off parameter of my non model obtained through training a neural network. At this stage, my argument is due to the fact that Bayesian nonparameteric estimation offers some insight on how to proceed on this problem. As I’ve been writing here for awhile, I’ve been interested in weakly informative priors. But I have little experience with nonparametric models. Perhaps Aki Vehtari or David Dunson or some other expert on these models can discuss how to set them up with weakly informative priors? This sounds like it could be important to me.
2 0.95972538 1475 andrew gelman stats-2012-08-30-A Stan is Born
Introduction: Stan 1.0.0 and RStan 1.0.0 It’s official. The Stan Development Team is happy to announce the first stable versions of Stan and RStan. What is (R)Stan? Stan is an open-source package for obtaining Bayesian inference using the No-U-Turn sampler, a variant of Hamiltonian Monte Carlo. It’s sort of like BUGS, but with a different language for expressing models and a different sampler for sampling from their posteriors. RStan is the R interface to Stan. Stan Home Page Stan’s home page is: http://mc-stan.org/ It links everything you need to get started running Stan from the command line, from R, or from C++, including full step-by-step install instructions, a detailed user’s guide and reference manual for the modeling language, and tested ports of most of the BUGS examples. Peruse the Manual If you’d like to learn more, the Stan User’s Guide and Reference Manual is the place to start.
3 0.95811892 1365 andrew gelman stats-2012-06-04-Question 25 of my final exam for Design and Analysis of Sample Surveys
Introduction: 25. You are using multilevel regression and poststratification (MRP) to a survey of 1500 people to estimate support for the space program, by state. The model is fit using, as a state- level predictor, the Republican presidential vote in the state, which turns out to have a low correlation with support for the space program. Which of the following statements are basically true? (Indicate all that apply.) (a) For small states, the MRP estimates will be determined almost entirely by the demo- graphic characteristics of the respondents in the sample from that state. (b) For small states, the MRP estimates will be determined almost entirely by the demographic characteristics of the population in that state. (c) Adding a predictor specifically for this model (for example, a measure of per-capita space-program spending in the state) could dramatically improve the estimates of state-level opinion. (d) It would not be appropriate to add a predictor such as per-capita space-program spen
4 0.95348418 1197 andrew gelman stats-2012-03-04-“All Models are Right, Most are Useless”
Introduction: The above is the title of a talk that Thad Tarpey gave at the Joint Statistical Meetings in 2009. Here’s the abstract: Students of statistics are often introduced to George Box’s famous quote: “all models are wrong, some are useful.” In this talk I [Tarpey] argue that this quote, although useful, is wrong. A different and more positive perspective is to acknowledge that a model is simply a means of extracting information of interest from data. The truth is infinitely complex and a model is merely an approximation to the truth. If the approximation is poor or misleading, then the model is useless. In this talk I give examples of correct models that are not true models. I illustrate how the notion of a “wrong” model can lead to wrong conclusions. I’m curious what he had to say—maybe he could post the slides? P.S. And here they are !
5 0.95083177 2146 andrew gelman stats-2013-12-24-NYT version of birthday graph
Introduction: They didn’t have room for all four graphs of the time-series decomposition so they just displayed the date-of-year graph: They rotated so the graph fit better on the page. The rotation worked for me, but I was a bit bummed that that they put the title and heading of the graph (“The birthrate tends to drop on holidays . . .”) on the left in the Mar-Apr slot, leaving no room to label Leap Day and April Fool’s. I suggested to the graphics people that they put the label at the very top and just shrink the rest of the graph by 5 or 10% so as to not take up any more total space. Then there’d be plenty of space to label Leap Day and April Fool’s. But they didn’t do it, maybe they felt that it wouldn’t look good to have the label right at the top, I dunno.
6 0.94908869 671 andrew gelman stats-2011-04-20-One more time-use graph
7 0.94345975 1426 andrew gelman stats-2012-07-23-Special effects
9 0.93975437 1021 andrew gelman stats-2011-11-21-Don’t judge a book by its title
10 0.93475604 2074 andrew gelman stats-2013-10-23-Can’t Stop Won’t Stop Mister P Beatdown
11 0.93363065 2062 andrew gelman stats-2013-10-15-Last word on Mister P (for now)
12 0.93351626 2056 andrew gelman stats-2013-10-09-Mister P: What’s its secret sauce?
13 0.93036568 1367 andrew gelman stats-2012-06-05-Question 26 of my final exam for Design and Analysis of Sample Surveys
14 0.9280715 758 andrew gelman stats-2011-06-11-Hey, good news! Your p-value just passed the 0.05 threshold!
15 0.92264366 2161 andrew gelman stats-2014-01-07-My recent debugging experience
16 0.91815561 2055 andrew gelman stats-2013-10-08-A Bayesian approach for peer-review panels? and a speculation about Bruno Frey
17 0.91556013 1811 andrew gelman stats-2013-04-18-Psychology experiments to understand what’s going on with data graphics?
18 0.91056502 1240 andrew gelman stats-2012-04-02-Blogads update
19 0.90885592 2061 andrew gelman stats-2013-10-14-More on Mister P and how it does what it does
20 0.90778214 1753 andrew gelman stats-2013-03-06-Stan 1.2.0 and RStan 1.2.0