Search

Quite frequently, catastrophes impact populated areas of the world, and hence the need for proper risk evaluations that support mitigation and management processes. Because of the uncertain nature of extreme natural hazards and lack of data, forecasts of the potential damage and losses before the event happens are needed. Catastrophe (CAT) models build on scenarios that represent all possible realizations of the hazard in terms of recurrence and intensity. Probabilistic risk models require the characterization of the hazards, the exposure model for the infrastructure, and its vulnerability. The main objective of this research is to compare loss exceedance curves, probable maximum loss curves, and average annual losses using four different available seismic hazard models for Chile. To isolate the effect of changing the hazard model in the risk results, the exposure and vulnerability information is fixed to the one available from the Global Assessment Report, GAR 15, and GAR ATLAS 2017. Imprecise probability theory, logic trees, and frequency and severity blends used by CAT modelers are the approaches applied and compared herein to propose either model blending or an interval of possible realizations. Both types of results have pros and cons. Blended results are point estimates which make them useful in a more traditional way, but their computation necessarily implies assigning weights to the models according to the modeler preferences. On the other hand, raw intervals of variability (without any knowledge of how the variable is distributed inside) are more transparent, as they simply state the bound of what is known without any preference, but their use is less understood among practitioners and could be even impractical.

For the reliability analysis with imprecise probability distributions, the failure probability and its sensitivity are always denoted as the functions of distribution parameters. The computation of those functional relationships remains a challenging task due to the intensive computational burden. To further improve the computational efficiency, this work proposes a new computational method based on the classical active learning Kriging model with U learning function. The Kriging model is constructed with a group of initial point and updated by the best contributing samples selected from different sample spaces corresponding to discrete distribution parameters, which guarantees the prediction ability in the variation range of distribution parameters. Consequently, the functions of failure probability and sensitivity can be estimated by the same Kriging model, which avoids re-constructing Kriging models corresponding to different distribution parameters. Several examples including two numerical examples and three engineering practices are investigated to validate the reasonability and superiority of the proposed method. Finally, the proposed method is applied to the fatigue life reliability and sensitivity analyses of a turbine disk structure.

In this paper we extend the notion of stochastic efficiency and inefficiency in portfolio optimization to the case of incomplete information by means of set-valued probabilities. The notion of set-valued probability models the concept of incomplete information about the underlying probability space and the probability associated with each scenario. Unlike other approaches in literature, our notion of inefficiency is introduced by means of the Monge–Kantorovich metric. We provide some numerical examples to illustrate this approach.

Conditional mutual information (CMI) maximization is a promising criterion for feature selection in a computationally efficient stepwise way, but it is hard to be applied comprehensively because of imprecise probability calculation and heavy computational load. Many dimension-reduced CMI-based and mutual information (MI)-based methods have been reported to achieve state-of-art performances in terms of classification. However, model deviations are introduced into the CMI and MI formulations in these methods during dimension reduction. In this paper, we start with the full-dimensional CMI to deal with the feature selection problem, so as to retain full inter-feature and feature-label mutual information when selecting new features. The cost function is approximated and simplified from a mathematical perspective to overcome the difficulties for maximizing the original full-dimensional CMI. A relationship is established between the proposed feature selection criterion and the one based on Hilbert-Schmidt independence, which explains qualitatively how the new criterion succeeds to achieve relevance maximization and redundance minimization simultaneously. Experiments on real-world datasets demonstrate the predominance of the proposed method over the existing ones.

Recent advances in decision making have incorporated both risk and ambiguity in decision theory and optimization methods. These methods implement a variety of uncertainty representations from probabilistic and non-probabilistic foundations, including traditional probability theory, sets of probability measures, uncertainty sets, ambiguity sets, possibility theory, evidence theory, fuzzy measures, and imprecise probability. The choice of uncertainty representation impacts the expressiveness and tractability of the decision models. We survey recent approaches for representing uncertainty in both decision making and optimization to clarify the trade-offs among the alternative representations. Robust and distributionally robust optimization are surveyed, with particular attention to standard form ambiguity sets. Applications of uncertainty and decision models are also reviewed, with a focus on recent optimization applications. These applications highlight common practices and potential research gaps. The intersection of behavioral decision making and robust optimization is a promising area for future research and there is also opportunity for further advances in distributionally robust optimization in sequential and multi-agent settings.

We present a new nonparametric predictive inference (NPI) method using a power-normal model for accelerated life testing (ALT). Combined with the accelerating link function and imprecise probability theory, the proposed method is a feasible way to predict the life of the product using ALT failure data. To validate the method, we run a series of simulations and conduct accelerated life tests with real products. The NPI lower and upper survival functions show the robustness of our method for life prediction. This is a continuous research, and some progresses have been made by updating the link function between different stress levels. We also explain how to renew and apply our model. Moreover, discussions have been made about the performance.

The factors that influence the inherent reliability of machine tool spindle are a mixture of various uncertainties and it leads that the reliability modeling and analysis of machine tool spindle can’t be dealt with by one mathematical theory. Meanwhile, the reliability data of the machine tool spindle for reliability modeling and analysis is often insufficient, and data of different types such as accumulated historical data, expert opinions, simulation data, etc. are used to make up for the lack of data. Thus, the unified quantification of mixed uncertainties and the data characterization of different types are the major premises for reliability modeling and analysis of machine tool spindle. By considering this, this paper makes use of the advantage of imprecise probability theory in quantizing the multiple types of data and the advantage of the Bayes theory in data fusion, and proposes a new inherent reliability modeling and analysis method based on imprecise Dirichlet model. In the proposed method, imprecise probability theory is used to quantify mixed uncertainties, imprecise Dirichlet model is built to characterize the different types of reliability data. After analyzing the inherent reliability variation regularity, an inherent reliability model is built, and the proposed method is verified by the inherent reliability calculation of a certain heavy-duty CNC machine tool’s milling spindle. This study can provide new method, theory and reference for reliability modeling and analysis when there are various uncertainties mixed and multiple data existed.

We provide counterexamples to some purported characterizations of dilation due to Pedersen and Wheeler (Erkenntnis 79(6):1305–1342, 2014, ISIPTA ’15: Proceedings of the 9th international symposium on imprecise probability: theories and applications, 2015).

With the tremendous development of financial institutions, credit risk prediction (CRP) plays an essential role in granting loans to customers and helps them to minimize their loss because credit approval sometimes results in massive financial loss. So extra attention is needed to identify risky customer. Researchers have designed complex CRP models using artificial intelligence (AI) and statistical techniques to support the financial institutions to take correct business decisions. Though there are various statistical and AI methods available, the recent literature shows that the ensemble-based CRP model provides improved prediction results than single classifier system. The small increase in the performance of CRP model could result in a significant improvement in the profit of financial institutions and banks. This work proposes a weight-adjusted boosting ensemble method (WABEM) using rough set (RS)-based feature selection (FS) technique with the balancing and regression-based preprocessing called RS$$\_$$RFS-WABEM. Regression is used to fill missing value in the records to improve the performance of CRP. Three credit datasets (Australia, German and Japanese) are chosen to validate the feasibility and effectiveness of the proposed ensemble method. The trade-off between the uncertainty and imprecise probability of the proposed classifier model is evaluated using the performance measures such as accuracy and area under the curve. Experimental results show that the proposed ensemble method performs better than other base and ensemble classifier methods.

Recent work has formally linked the traditional axiomatisation of incomplete preferences a la Anscombe-Aumann with the theory of desirability developed in the context of imprecise probability, by showing in particular that they are the very same theory. The equivalence has been established under the constraint that the set of possible prizes is finite. In this paper, we relax such a constraint, thus de facto creating one of the most general theories of rationality and decision making available today. We provide the theory with a sound interpretation and with basic notions, and results, for the separation of beliefs and values, and for the case of complete preferences. Moreover, we discuss the role of conglomerability for the presented theory, arguing that it should be a rationality requirement under very broad conditions.

Bayesian decision theory was invented by Leonard Savage, who is on record as saying that it would be “preposterous” and “utterly ridiculous” to apply his theory except in a small world. But modern Bayesians proceed as though Savage’s theory is always the rational way to make choices in all circumstances. What is a small world? Why did Savage restrict his theory to small worlds? Where did Savage think priors come from? What are the implications for behavioral applications? How could Savage’s theory be generalized to apply to at least some large worlds? This paper offers some partial answers to such questions.

In almost all the realistic circumstances, such as atmospheric dispersion and health risk assessment analysis, it is very essential to include all the information into modelling. The parameters associated with a particular model may include different kinds of variability, imprecision and uncertainty. More often, it is seen that available information is interpreted in probabilistic sense. Probability theory is a well-established theory to measure such kind of variability. However, not all available information, data or model parameters affected by variability, imprecision and uncertainty can be handled by traditional probability theory. Uncertainty or imprecision may occur due to incomplete information or data, measurement error or data obtained from expert judgment or subjective interpretation of available data or information. Thus, model parameters’ data may be affected by subjective uncertainty. Traditional probability theory is inappropriate to represent subjective uncertainty. Possibility theory is another branch of mathematics which is used as a tool to describe the parameters with insufficient knowledge. It may also happen that the probabilistic information about the parameters may be imprecise; that is, it might be associated with some probability bounds. Such types of uncertainty are incorporated into imprecise probability, a theory that extends traditional probability theory by allowing for intervals or sets of probabilities. In this paper, an attempt has been made to combine imprecise probability knowledge and possibility knowledge and draw the uncertainty. The article describes a combination of imprecise probability and fuzzy knowledge and applied to atmospheric dispersion with a case study. Further, fuzziness measure has been utilized in quantifying the uncertainty.

International audience; Within imprecise probability theory, the extreme points of convex probability sets have an important practical role (to perform inference on graphical models, to compute expectation bounds,. . .). This is especially true for sets presenting specic features that make them easy to manipulate in applications. This easiness is the reason why extreme points of such models (probability intervals, possibility distributions,. . .) have been well studied. Yet, imprecise cumulative distributions (a.k.a. p-boxes) constitute an important exception, as the characterization of their extreme points remain to be studied. This is what we do in this paper, where we characterize the maximal number of extreme points of a p-box, give a family of p-boxes that attains this number and show an algorithm that allows to compute the extreme points of a given p-box. To achieve all this, we also provide what we think to be a new characterization of extreme points of a belief function.

In the design and manufacturing of mechanical components, the dynamic properties of continuum structure are one of the most significant performances. At the same time, the uncertainty is widespread in these dynamic problems. This paper presents a robust topology optimization methodology of structure for dynamic properties with consideration of hybrid uncertain parameters. The imprecise probability uncertainties including materials, geometry and boundary condition are treated as an interval random model, in which the probability distribution parameters of random variables are modeled as the interval variables instead of given precise values. Two dynamic properties, including dynamic-compliance and eigenvalue, are chosen as the objective function. In addition, different excitation frequency or eigenvalue is discussed. In this work, the bi-directional evolutionary structural optimization (BESO) method is adopted to find the optimal robust layout of the structure. A series of numerical examples is presented to illustrate the optimization procedure, and the effectiveness of the proposed method is demonstrated clearly.

Small sample size problem is one of the main problems that heavy numerical control (NC) machine tools encounter in their reliability assessment. In order to deal with the small sample size problem, many indirect reliability data such as reliability data of similar products, expert opinion, and engineers’ experience are used in reliability assessment. However, the existing mathematical theories cannot simultaneously process the above reliability data of multiple types, and thus imprecise probability theory is introduced. Imprecise probability theory can simultaneously process multiple reliability data by quantifying multiple uncertainties (stochastic uncertainty, fuzzy uncertainty, epistemic uncertainty, etc.) together. Although imprecise probability theory has so many advantages, the existing natural extension models are complex and the computation result is imprecise. Therefore, they need some improvement for the better application of reliability engineering. This paper proposes an improved imprecise reliability assessment method by introducing empirical probability distributions to natural extension model, and the improved natural extension model is applied to the reliability assessment of heavy NC machine tool spindle to illustrate its effectiveness.

Markowitz’s work has had a major impact on academic research and the financial industry as a whole. The main idea of his model is risk aversion of average investors and their desire to maximise the expected return with the least risk. In this paper we extend the classical Markowitz’s model by introducing a portfolio optmization model in which the underlying space of events is described in terms of a probability multimeasure. The notion of probability multimeasure allows to formalize the concept of imprecise probability measure and incomplete information.

We describe the Dempster---Shafer belief structure and provide some of its basic properties. We introduce the plausibility and belief measures associated with a belief structure. We note that these are not the only measures that can be associated with a belief structure. We describe a general approach for generating a class of measures that can be associated with a belief structure using a monotonic function on the unit interval, called a weight generating function. We study a number of these functions and the measures that result. We show how to use weight-generating functions to obtain dual measures from a belief structure. We show the role of belief structures in representing imprecise probability distributions. We describe the use of dual measures, other then plausibility and belief, to provide alternative bounding intervals for the imprecise probabilities associated with a belief structure. We investigate the problem of decision making under belief structure type uncertain. We discuss two approaches to this decision problem. One of which is based on an expected value of the OWA aggregation of the payoffs associated with the focal elements. The second approach is based on using the Choquet integral of a measure generated from the belief structure. We show the equivalence of these approaches.

Quantum mechanics and probability theory share one peculiarity. Both have well established mathematical formalisms, yet both are subject to controversy about the meaning and interpretation of their basic concepts. Since probability plays a fundamental role in QM, the conceptual problems of one theory can affect the other. We first classify the interpretations of probability into three major classes: (a) inferential probability, (b) ensemble probability, and (c) propensity. Class (a) is the basis of inductive logic; (b) deals with the frequencies of events in repeatable experiments; (c) describes a form of causality that is weaker than determinism. An important, but neglected, paper by P. Humphreys demonstrated that propensity must differ mathematically, as well as conceptually, from probability, but he did not develop a theory of propensity. Such a theory is developed in this paper. Propensity theory shares many, but not all, of the axioms of probability theory. As a consequence, propensity supports the Law of Large Numbers from probability theory, but does not support Bayes theorem. Although there are particular problems within QM to which any of the classes of probability may be applied, it is argued that the intrinsic quantum probabilities (calculated from a state vector or density matrix) are most naturally interpreted as quantum propensities. This does not alter the familiar statistical interpretation of QM. But the interpretation of quantum states as representing knowledge is untenable. Examples show that a density matrix fails to represent knowledge.

The concept of probability seems to have been inexplicable since its invention in the seventeenth century. In its use in science, probability is closely related with relative frequency. So the task seems to be interpreting that relation. In this paper, we start with predicted relative frequency and show that its structure is the same as that of probability. I propose to call that the ‘prediction interpretation’ of probability. The consequences of that definition are discussed. The “ladder”-structure of the probability calculus is analyzed. The expectation of the relative frequency is shown to be equal to the predicted relative frequency. Probability is shown to be the most general empirically testable prediction.

Understanding probabilities as something other than point values (e.g., as intervals) has often been motivated by the need to find more realistic models for degree of belief, and in particular the idea that degree of belief should have an objective basis in “statistical knowledge of the world.” I offer here another motivation growing out of efforts to understand how chance evolves as a function of time. If the world is “chancy” in that there are non-trivial, objective, physical probabilities at the macro-level, then the chance of an event e that happens at a given time is $$

The foundations of probability are viewed through the lens of the subjectivist interpretation. This article surveys conditional probability, arguments for probabilism, probability dynamics, and the evidential and subjective interpretations of probability.

The proposed minimax procedure blends strict Bayesian methods with p values and confidence intervals or with default-prior methods. Two applications to hypothesis testing bring some implications to light. First, the blended probability that a point null hypothesis is true is equal to the p value or a lower bound of an unknown posterior probability, whichever is greater. As a result, the p value is reported instead of any posterior probability in the case of complete prior ignorance but is ignored in the case of a fully known prior. In the case of partial knowledge about the prior, the possible posterior probability that is closest to the p value is used for inference. The second application provides guidance on the choice of methods used for small numbers of tests as opposed to those appropriate for large numbers. Whereas statisticians tend to prefer a multiple comparison procedure that adjusts each p value for small numbers of tests, large numbers instead lead many to estimate the local false discovery rate (LFDR), a posterior probability of hypothesis truth. Each blended probability reduces to the LFDR estimate if it can be estimated with sufficient accuracy or to the adjusted p value otherwise.

When developing a species distribution model, usually one tests several competing models such as logistic regressions characterized by different sets of covariates. Yet, there is an exponential number of subsets of covariates to choose from. This generates the problem of model uncertainty. Bayesian model averaging (BMA) is a state-of-the-art approach to deal with model uncertainty. BMA weights the inferences of multiple models. However, the results yielded by BMA depend on the prior probability assigned to the models. Credal model averaging (CMA) extends BMA towards robustness. It substitutes the single prior over the models by a set of priors. The CMA inferences (e.g., posterior probability of inclusion of a covariate, coefficient of a covariate, probability of presence) are intervals. The interval shows the sensitivity of the BMA estimate on the prior over the models. CMA detects the prior-dependent instances, namely cases in which the most probable outcome becomes presence or absence depending on the adopted prior over the models. On such prior-dependent instances, BMA behaves almost as a random guesser. The weakness of BMA on the prior-dependent instances is to our knowledge pointed out for the first time in the ecological literature. On the prior-dependent instances CMA avoids random guessing acknowledging undecidability. In this way it stimulates the decision maker to convey further information before taking the decision. We provide thorough experiments on different data sets.

This article presents a new statistical inference method for classification. Instead of minimizing a loss function that solely takes residuals into account, it uses the Kolmogorov–Smirnov bounds for the cumulative distribution function of the residuals, as such taking conservative bounds for the underlying probability distribution for the population of residuals into account. The loss functions considered are based on the theory of support vector machines. Parameters for the discriminant functions are computed using a minimax criterion, and for a wide range of popular loss functions, the computations are shown to be feasible based on new optimization results presented in this article. The method is illustrated in examples, both with small simulated data sets and with real-world data.

Dilation occurs when an interval probability estimate of some event E is properly included in the interval probability estimate of E conditional on every event F of some partition, which means that one’s initial estimate of E becomes less precise no matter how an experiment turns out. Critics maintain that dilation is a pathological feature of imprecise probability models, while others have thought the problem is with Bayesian updating. However, two points are often overlooked: (1) knowing that E is stochastically independent of F (for all F in a partition of the underlying state space) is sufficient to avoid dilation, but (2) stochastic independence is not the only independence concept at play within imprecise probability models. In this paper we give a simple characterization of dilation formulated in terms of deviation from stochastic independence, propose a measure of dilation, and distinguish between proper and improper dilation. Through this we revisit the most sensational examples of dilation, which play up independence between dilator and dilatee, and find the sensationalism undermined by either fallacious reasoning with imprecise probabilities or improperly constructed imprecise probability models.

For the imprecise probability distribution of structural system, the variance based importance measures (IMs) of the inputs are investigated, and three IMs are defined on the conditions of random distribution parameters, interval distribution parameters and the mixture of those two types of distribution parameters. The defined IMs can reflect the influence of the inputs on the output of the structural system with imprecise distribution parameters, respectively. Due to the large computational cost of the variance based IMs, sparse grid method is employed in this work to compute the variance based IMs at each reference point of distribution parameters. For the three imprecise distribution parameter cases, the sparse grid method and the combination of sparse grid method with genetic algorithm are used to compute the defined IMs. Numerical and engineering examples are employed to demonstrate the rationality of the defined IMs and the efficiency of the applied methods.

Probability measures are properly considered subjective, if only as a result of the knowledge limitations of human experience. Even if objective probability can be demonstrated as an unambiguously verifiable property of the physical world, our necessarily imperfect access to such aspects of reality would necessitate the construction of subjective probability measures for examinations of human behavior. This paper will explore the distinction between subjective and purportedly objective probability in principle, and demonstrate that actual probabilities can only be subjective, at least ex ante. The paper will next examine the persistent efforts of philosophers, mathematicians, and particularly quantitatively-inclined social scientists, to construct and employ mistaken conceptions of purportedly objective probability. The neo-indeterminist interpretation of sub-atomic particle physics is addressed next, with its attempt to construct an objective-probabilistic account of reality based on the Heisenberg uncertainty principle. An explanation is suggested for the persistent allure of objective probability in the social sciences, which purports to offer a more economical and parsimonious basis for human behavior and the formation of expectations. The view that there are ontologically privileged, objectively-verifiable probabilities and expectations, will be examined and criticized. It appears to be one source of the discredited, though nonetheless widespread and persistent, belief in the efficacy and feasibility of central economic planning.

We resolve a useful formulation of the question how a statistician can coherently incorporate the information in a consulted expert’s probability assessment for an event into a personal posterior probability assertion. Using a framework that recognises the total information available as composed of units available only to each of them along with units available to both, we show: that a sufficient statistic for all the information available to both the expert and the statistician is the product of their odds ratios in favour of the event; that the geometric mean of their two probabilities specifies a contour of pairs of assertions in the unit-square that yield the same posterior probability; that the information-combining function is parameterised by an unknown probability for the event conditioned only on the unspecified information common to both the statistician and the expert; and that an assessable mixing distribution over this unspecified probability allows an integrable mixture distribution to represent a computable posterior probability. The exact results allow the identification of the subclass of coherent probabilities that are externally Bayesian operators. This subclass is equivalent to the class of combining functions that honour the principles of uniformity and compromise.

In conditional probabilistic logic programming, given a query, the two most common forms for answering the query are either a probability interval or a precise probability obtained by using the maximum entropy principle. The former can be noninformative (e.g., interval [0, 1]) and the reliability of the latter is questionable when the priori knowledge is imprecise. To address this problem, in this paper, we propose some methods to quantitatively measure if a probability interval or a single probability is sufficient for answering a query. We first propose an approach to measuring the ignorance of a probabilistic logic program with respect to a query. The measure of ignorance (w.r.t. a query) reflects how reliable a precise probability for the query can be and a high value of ignorance suggests that a single probability is not suitable for the query. We then propose a method to measure the probability that the exact probability of a query falls in a given interval, e.g., a second order probability. We call it the degree of satisfaction. If the degree of satisfaction is high enough w.r.t. the query, then the given interval can be accepted as the answer to the query. We also prove our measures satisfy many properties and we use a case study to demonstrate the significance of the measures.

In shield tunneling, the control system needs very reliable capability of deviation rectifying in order to ensure that the tunnel trajectory meets the permissible criterion. To this goal, we present an approach that adopts Markov decision process (MDP) theory to plan the driving force with explicit representation of the uncertainty during excavation. The shield attitudes of possible world and driving forces during excavation are scattered as a state set and an action set, respectively. In particular, an evaluation function is proposed with consideration of the stability of driving force and the deviation of shield attitude. Unlike the deterministic approach, the driving forces based on MDP model lead to an uncertain effect and the attitude is known only with an imprecise probability. We consider the case that the transition probability varies in a given domain estimated by field data, and discuss the optimal policy based on the interval arithmetic. The validity of the approach is discussed by comparing the driving force planning with the actual operating data from the field records of Line 9 in Tianjin. It is proved that the MDP model is reasonable enough to predict the driving force for automatic deviation rectifying.

We detail the relationship between sets of desirable gambles and conditional lower previsions. The former is one the most general models of uncertainty. The latter corresponds to Walley's celebrated theory of imprecise probability. We consider two avenues: when a collection of conditional lower previsions is derived from a set of desirable gambles, and its converse. In either case, we relate the properties of the derived model with those of the originating one. Our results constitute basic tools to move from one formalism to the other, and thus to take advantage of work done in the two fronts.

A fuzzy expected value of the possibility-probability distribution is a set with \(\underline{E}_{\alpha}(x)\) and \(\overline{E}_{\alpha}(x)\) as its boundaries. The fuzzy expected values \(\underline{E}_{\alpha}(x)\) and \(\overline{E}_{\alpha}(x)\) of a possibility-probability distribution represent the fuzzy risk values being calculated. Using these values under a given α level, three risk values can be calculated: a conservative risk value, a venture risk value, and a maximum probability risk value. Calculation of the fuzzy expected value of Jinhua City’s water resource risk has been performed based on the interior-exterior set model. This model is first used to evaluate the risk of water resources in Jinhua City: it not only solves an imprecise probability estimation, which results from small samples and unclear risk relationship, but it also explores the implicit risk information of the raw data as much as possible. Both of these achievements can make analyses more objective and comprehensive, which makes it easy to regulate options for policy-makers. Hence, the fuzzy risk analysis provides a new way to assess water resources.

The thesis of this article is that the nature of probability is centered on its formal properties, not on any of its standard interpretations. Section 2 is a survey of Bayesian applications. Section 3 focuses on two examples from physics that seem as completely objective as other physical concepts. Section 4 compares the conflict between subjective Bayesians and objectivists about probability to the earlier strident conflict in physics about the nature of force. Section 5 outlines a pragmatic approach to the various interpretations of probability. Finally, Sect. 6 argues that the essential formal nature of probability is expressed in the standard axioms, but more explicit attention should be given to the concept of randomness.

In traditional Bayesian software reliability models, it was assume that all probabilities are precise. In practical applications the parameters of the probability distributions are often under uncertainty due to strong dependence on subjective information of experts’ judgments on sparse statistical data. In this paper, a quasi-Bayesian software reliability model using interval-valued probabilities to clearly quantify experts’ prior beliefs on possible intervals of the parameters of the probability distributions is presented. The model integrates experts’ judgments with statistical data to obtain more convincible assessments of software reliability with small samples. For some actual data sets, the presented model yields better predictions than the Jelinski-Moranda (JM) model using maximum likelihood (ML).

This paper explores the application of semi-qualitative probabilistic networks (SQPNs) that combine numeric and qualitative information to computer vision problems. Our version of SQPN allows qualitative influences and imprecise probability measures using intervals. We describe an Imprecise Dirichlet model for parameter learning and an iterative algorithm for evaluating posterior probabilities, maximum a posteriori and most probable explanations. Experiments on facial expression recognition and image segmentation problems are performed using real data.

On grounds of insufficient reason, a probability of 1/K is assigned to K mutually exclusive possible events when nothing is known about the likelihood of those events. Fox and Rottenstreich (2003) present evidence that subjective probability judgments are typically biased towards this ignorance prior, and therefore depend on the partition K. Results from two studies indicate that lower-upper (imprecise) probability judgments by naive judges also exhibit partition dependence, despite the potential that imprecise probabilities provide for avoiding it. However, beta regression reveals two kinds of priming effects, one of which is modeled by mixture distributions. Another novel finding suggests that when partition primes conflict with a normatively correct partition some judges widen their probability intervals to encompass both partitions. The results indicate that imprecise probability judgments may be better suited than precise probabilities for handling conflicting or ambiguous information about ...

This paper considers the problem of choosing between an existing component whose reliability is well established and a new component that has an unknown reliability. In some scenarios, the designer may have some initial beliefs about the new component’s reliability. The designer may also have the opportunity to obtain more information and to update these beliefs. Then, based on these updated beliefs, the designer must make a decision between the two components. This paper examines the statistical approaches for updating reliability assessments and the decision policy that the designer uses. We consider four statistical approaches for modeling the uncertainty about the new component and updating assessments of its reliability: a classical approach, a precise Bayesian approach, a robust Bayesian approach, and an imprecise probability approach. The paper investigates the impact of different approaches on the decision between the components and compares them. In particular, given that the test results are random, the paper considers the likelihood of making a correct decision with each statistical approach under different scenarios of available information and true reliability. In this way, the emphasis is on practical comparisons of the policies rather than on philosophical arguments.

Our main aim in this paper is to inform the physics community (and especially experts in quantum information) about investigations of the problem of the probabilistic compatibility of a family of random variables: the possibility of realizing such a family based on a single probability measure (of constructing a single Kolmogorov probability space). These investigations were started a hundred years ago by Boole. The complete solution of the problem was obtained by the Soviet mathematician Vorobiev in the 1960s. It turns out that probabilists and statisticians obtained inequalities for probabilities and correlations that include the famous Bell’s inequality and its generalizations.

In the paper, we consider the application of the theory of probability metrics in several areas in the field of finance. First, we argue that specially structured probability metrics can be used to quantify stochastic dominance relations. Second, the methods of the theory of probability metrics can be used to arrive at a general axiomatic treatment of dispersion measures and probability metrics can be used to describe continuity of risk measures. Finally, the methods of probability metrics theory are applied to the benchmark-tracking problem significantly extending the problem setting.

The right interpretation of subjective probability is implicit in the theories of upper and lower odds, and upper and lower previsions, developed, respectively, by Cedric Smith (1961) and Peter Walley (1991). On this interpretation you are free to assign contingent events the probability 1 (and thus to employ conditionalization as a method of probability revision) without becoming vulnerable to a weak Dutch book.

In the past, arithmetic and geometric means have both been used to characterise pathogen densities in samples used for microbial risk assessment models. The calculation of total (annual) risk is based on cumulative independent (daily) exposures and the use of an exponential dose–response model, such as that used for exposure to Giardia or Cryptosporidium. Mathematical analysis suggests that the arithmetic mean is the appropriate measure of central tendency for microbial concentration with respect to repeated samples of daily exposure in risk assessment. This is despite frequent characterisation of microbial density by the geometric mean, since the microbial distributions may be Log normal or skewed in nature. Mathematical derivation supporting the use of the arithmetic mean has been based on deterministic analysis, prior assumptions and definitions, the use of point-estimates of probability, and has not included from the outset the influence of an actual distribution for microbial densities. We address these issues by experiments using two real-world pathogen datasets, together with Monte Carlo simulation, and it is revealed that the arithmetic mean also holds in the case of a daily dose with a finite distribution in microbial density, even when the distribution is very highly-skewed, as often occurs in environmental samples. Further, for simplicity, in many risk assessment models, the daily infection risk is assumed to be the same for each day of the year and is represented by a single value, $$ \hat{p}, $$ which is then used in the calculation of p Σ, which is a numerical estimate of annual risk, P Σ, and we highlight the fact that $$ \hat{p} $$ is simply a function of the geometric mean of the daily complementary risk probabilities (although it is sometimes approximated by the arithmetic mean of daily risk in the low dose case). Finally, the risk estimate is an imprecise probability with no indication of error and we investigate and clarify the distinction between risk and uncertainty assessment with respect to the predictive model used for total risk assessment.

We derive an implementable algorithm for solving nonlinear stochastic optimization problems with failure probability constraints using sample average approximations. The paper extends prior results dealing with a failure probability expressed by a single measure to the case of failure probability expressed in terms of multiple performance measures. We also present a new formula for the failure probability gradient. A numerical example addressing the optimal design of a reinforced concrete highway bridge illustrates the algorithm.

The reference class problem arises when we want to assign a probability to a proposition (or sentence, or event) X, which may be classified in various ways, yet its probability can change depending on how it is classified. The problem is usually regarded as one specifically for the frequentist interpretation of probability and is often considered fatal to it. I argue that versions of the classical, logical, propensity and subjectivist interpretations also fall prey to their own variants of the reference class problem. Other versions of these interpretations apparently evade the problem. But I contend that they are all "no-theory" theories of probability - accounts that leave quite obscure why probability should function as a guide to life, a suitable basis for rational inference and action. The reference class problem besets those theories that are genuinely informative and that plausibly constrain our inductive reasonings and decisions. I distinguish a "metaphysical" and an "epistemological" reference class problem. I submit that we can dissolve the former problem by recognizing that probability is fundamentally a two-place notion: conditional probability is the proper primitive of probability theory. However, I concede that the epistemological problem remains.

This article presents the results of a survey designed to test, with economically sophisticated participants, Ellsberg’s ambiguity aversion hypothesis, and Smithson’s conflict aversion hypothesis. Based on an original sample of 78 professional actuaries (all members of the French Institute of Actuaries), this article provides empirical evidence that ambiguity (i.e. uncertainty about the probability) affect insurers’ decision on pricing insurance. It first reveals that premiums are significantly higher for risks when there is ambiguity regarding the probability of the loss. Second, it shows that insurers are sensitive to sources of ambiguity. The participants indeed, charged a higher premium when ambiguity came from conflict and disagreement regarding the probability of the loss than when ambiguity came from imprecision (imprecise forecast about the probability of the loss). This research thus documents the presence of both ambiguity aversion and conflict aversion in the field of insurance, and discuses economic and psychological rationales for the observed behaviours.

A non-monotonic theory of probability is put forward and shown to have applicability in the quantum domain. It is obtained simply by replacing Kolmogorov’s positivity axiom, which places the lower bound for probabilities at zero, with an axiom that reduces that lower bound to minus one. Kolmogorov’s theory of probability is monotonic, meaning that the probability of A is less then or equal to that of B whenever A entails B. The new theory violates monotonicity, as its name suggests; yet, many standard theorems are also theorems of the new theory since Kolmogorov’s other axioms are retained. What is of particular interest is that the new theory can accommodate quantum phenomena (photon polarization experiments) while preserving Boolean operations, unlike Kolmogorov’s theory. Although non-standard notions of probability have been discussed extensively in the physics literature, they have received very little attention in the philosophical literature. One likely explanation for that difference is that their applicability is typically demonstrated in esoteric settings that involve technical complications. That barrier is effectively removed for non-monotonic probability theory by providing it with a homely setting in the quantum domain. Although the initial steps taken in this paper are quite substantial, there is much else to be done, such as demonstrating the applicability of non-monotonic probability theory to other quantum systems and elaborating the interpretive framework that is provisionally put forward here. Such matters will be developed in other works.

Whilst the majority of the climate research community is now set upon the objective of generating probabilistic predictions of climate change, disconcerting reservations persist. Attempts to construct probability distributions over socio-economic scenarios are doggedly resisted. Variation between published probability distributions of climate sensitivity attests to incomplete knowledge of the prior distributions of critical parameters and structural uncertainties in climate models. In this paper we address these concerns by adopting an imprecise probability approach. We think of socio-economic scenarios as fuzzy linguistic constructs. Any precise emissions trajectory (which is required for climate modelling) can be thought of as having a degree of membership in a fuzzy scenario. Next, it is demonstrated how fuzzy scenarios can be propagated through a low-dimensional climate model, MAGICC. Fuzzy scenario uncertainties and imprecise probabilistic representation of climate model uncertainties are combined using random set theory to generate lower and upper cumulative probability distributions for Global Mean Temperature anomaly. Finally we illustrate how non-additive measures provide a flexible framework for aggregation of scenarios, which can represent some of the semantics of socio-economic scenarios that defy conventional probabilistic representation.

This note shows that if the space of events is sufficiently rich and the subjective probability function of each individual is non-atomic, then there is a σ-algebra of events over which everyone will have the same probability function, and moreover, the range of this common probability is the entire unit interval.

I focus my discussion on the well-known Ellsberg paradox. I find good normative reasons for incorporating non-precise belief, as represented by sets of probabilities, in an Ellsberg decision model. This amounts to forgoing the completeness axiom of expected utility theory. Provided that probability sets are interpreted as genuinely indeterminate belief (as opposed to “imprecise” belief), such a model can moreover make the “Ellsberg choices” rationally permissible. Without some further element to the story, however, the model does not explain how an agent may come to have unique preferences for each of the Ellsberg options. Levi (1986, Hard choices: Decision making under unresolved conflict. Cambridge, New York: Cambridge University Press) holds that the extra element amounts to innocuous secondary “risk” or security considerations that are used to break ties when more than one option is rationally permissible. While I think a lexical choice rule of this kind is very plausible, I argue that it involves a greater break with xpected utility theory than mere violation of the ordering axiom.