摘要 :
This paper reflects on a recent article by Heckman and Pinto (2013) in which they discuss a formal system, called do-calculus, that operationalizes Haavelmo's conception of policy intervention. They replace the do-operator with an...
展开
This paper reflects on a recent article by Heckman and Pinto (2013) in which they discuss a formal system, called do-calculus, that operationalizes Haavelmo's conception of policy intervention. They replace the do-operator with an equivalent operator called 'fix,' highlight the capabilities of 'fix,' discover limitations in 'do,' and inform readers that those limitations disappear in 'the Haavelmo approach.' I examine the logic of HP's paper, its factual basis, and its impact on econometric research and education.
收起
摘要 :
I commend Professor VanderWeele for providing a lucid description of the 'surrogate paradox' and, through it, a comprehensive discussion of the current state of thinking about surrogate endpoints, their function in experimental st...
展开
I commend Professor VanderWeele for providing a lucid description of the 'surrogate paradox' and, through it, a comprehensive discussion of the current state of thinking about surrogate endpoints, their function in experimental studies, and the various approaches devised to give them formal underpinnings. The first question that came to mind in reading VanderWeele's paper was: can we explain the phenomenon in simple terms, divorced from the technical vocabulary that was devised to formulate notions such as 'indirect effect,' 'principled strata,' 'proportion-mediated,' and perhaps others.`My second question was: If we take the negation of the 'surrogate paradox' as a criterion for 'good' surrogate, why can't we create a new, formal definition of 'surrogacy' that (1) will automatically avoid the paradox and (2) will settle, once for all, the disputes (among theoreticians) as to what 'approach' is best for defining surrogates (Joffe and Green, 2009, pp. 530-538; Pearl, 2011). In thinking about these two questions, I came across a simple way of explaining how the paradox comes about and, indirectly, why the requirement of avoiding the paradox could not, in itself, constitute a satisfactory definition of surrogacy.
收起
摘要 :
Types are a good starting point for various software reengineering tasks. Unfortunately, programs requiring reengineering most desperately are written in languages without an adequate type system (such as COBOL). To solve this pro...
展开
Types are a good starting point for various software reengineering tasks. Unfortunately, programs requiring reengineering most desperately are written in languages without an adequate type system (such as COBOL). To solve this problem, we propose a method of automated type inference for these languages. The main ingredients are that if variables are compared using some relational operator their types must be the same; likewise if an expression is assigned to a variable, the type of the expression must be a subtype of that of the variable. We present the formal type system and inference rules for this approach, show their effect on various real life COBOL fragments, describe the implementation of our ideas in a prototype type inference tool for COBOL, and discuss a number of applications.
收起
摘要 :
During this period of research a substantial research program continued instatistical inference, one that has seen the publication of nearly 100 papers and a book since the initiation of support. The papers written or revised in 1...
展开
During this period of research a substantial research program continued instatistical inference, one that has seen the publication of nearly 100 papers and a book since the initiation of support. The papers written or revised in 1991-1992 are: (1) An Assymptotic Theory for Sliced Inverse Regression , T. Hsing and R.J. Carroll (2) Measurement Error Regression with Unknown Link: Dimension Reduction and Data Visualization , Raymond Carroll and Ker-Chau Li (3) Further Remarks on Robustness in the Logistic Regression Model , R.J. Carroll and Shane Pederson (4) Semiparametric Comparison of Regression Curves Via Normal Likelihoods , R.J. Carroll and Peter Hall (5) Theoreticas Aspects of Ill-Posed Problems in Statistics , A.C.M. Van Rooij, F.H. Ruymgaart and R.J. Carroll.
收起
摘要 :
Geometric inference is widely used in computer vision, but very little attention has been given to the question of how geometric properties affect the resulting errors in the inferences made. This thesis addresses the problem of t...
展开
Geometric inference is widely used in computer vision, but very little attention has been given to the question of how geometric properties affect the resulting errors in the inferences made. This thesis addresses the problem of the stability of geometric inference in determining locations with a goal of being able to predict type and magnitude of the errors which occur and to determine on what basis to make geometric inferences which will minimize error. It is shown that the amount of the error occurring in a localization process using angular measurements to features depends heavily on which features are used, that the amount of the error occurring in such a localization process is not a function of the number of features used, that it is possible to develop simple heuristic functions for choosing features for localization which will significantly decrease error in that localization, that it is possible to decrease localization error in a particular direction, and that, if features have been identified but knowledge of left to right order in the view is unknown, simple steps can be taken to aid in determining that ordering. This knowledge is applied in the domain of robot navigation in outdoor, unstructured environments.
收起
摘要 :
Our research aims to develop computational models of inductive inference in higher-level human cognition, specifically the ability to generalize from sparse, noisy, ambiguous data, and to build more human-like machine learning sys...
展开
Our research aims to develop computational models of inductive inference in higher-level human cognition, specifically the ability to generalize from sparse, noisy, ambiguous data, and to build more human-like machine learning systems. Work has focused on two areas of cognition: learning about categories and their properties, and learning about relational structures -- specifically, systems of causal and social relations. We have developed a theory-based Bayesian framework for modeling these learning tasks as statistical inference over hierarchies of structured knowledge representations. Our models have made several contributions. First, they have explained a broad range of phenomena with high quantitative accuracy, using a minimum of free parameters. Second, they have provided a rational framework for explaining how and why everyday induction works, in terms of approximations to optimal statistical inference in natural environments. Third, they have provided tools for elucidating people's implicit theories about the structure of the world - describing the form and content of the prior knowledge that guides inductive inference and explaining how it may be acquired from experience. Finally, our models have led to improved algorithms for machine learning of category structures and their property distributions, causal networks, and the structure of social relations, thus bringing artificial intelligence systems closer to the capacities of human intelligence at the same tone as they support a better understanding of human intelligence.
收起
摘要 :
In survey sample inference, there are two fundamental positions that can be taken with respect to randomized designs and the use of design weights (inverses of selection probabilities): (1) both are necessary; (2) neither is neces...
展开
In survey sample inference, there are two fundamental positions that can be taken with respect to randomized designs and the use of design weights (inverses of selection probabilities): (1) both are necessary; (2) neither is necessary. Indeed, neither is necessary, and there are occasions when insistence on their use undermines inference. There are other occasions when, the analyst being at a remove from the sampling process, the selection probabilities are helpful information, which it makes sense to incorporate into the method of inference. Strict maximum likelihood inference (as distinguished from the pseudo-likelihood or weighted distribution likelihood approaches) can suitably incorporate the sample weights. The theory of this is not simple, but the practice usually is. We illustrate these points.
收起
摘要 :
There are a number of algorithms for calculating the least squares isotonic regression function and six of these algorithms are discussed in Section 2, 3 of Barlow, Bartholomew, Bremner and Brunk (1972). The most widely used of th...
展开
There are a number of algorithms for calculating the least squares isotonic regression function and six of these algorithms are discussed in Section 2, 3 of Barlow, Bartholomew, Bremner and Brunk (1972). The most widely used of these algorithms is the pool adjacent violators algorithm which is applicable only in the case of a simple linear ordering or an amalgamation of simple linear orderings. Cran (1980, Algorithm AS 149) gives a routine for computing the isotonic regression for the case of a simple linear ordering using the Up and Down Blocks algorithm. In many applications of isotonic regression we have more than one independent variable and the regression function is restricted to be monotone in each independent variable.
收起