Evidential Reasoning

Resolving domain incompatibility among independently developed databases often involves uncertain information. DeMichiel (1989) showed that uncertain information can be generated by the mapping of conflicting attributes to a common domain, based on some domain knowledge. The authors show that uncertain information can also arise when the database integration process requires information not directly represented in the component databases, but can be obtained through some summary of data. They therefore propose ...

Como un intento de responder a la pregunta que titula este trabajo, haré una breve reseña de los aspectos principales que particularizan al razonamiento probatorio judicial en relación con el realizado, con el mismo objeto, por fuera de los tribunales. ¿Qué tiene de especial? Según argumentaré, sus particularidades vienen dadas, mayormente, por el carácter reglado del contexto en el que se inserta, el proceso judicial. Las normas jurídicas influyen sobre el objeto de la indagación fáctica (qué clase de hechos son relevantes y cuáles objeto de prueba), con qué información se puede contar (según las reglas que rigen la actividad probatoria), las clases de razones que cuentan como justificación de las decisiones sobre los hechos probados, y los sujetos que toman esas decisiones. En las páginas que siguen me dedicaré a estudiar sucintamente estas influencias. La tarea tiene relevancia en el plano teórico puesto que, además de ayudar a comprender y modelar el fenómeno en sí, su resultado es una importante base de marcha para responder a otras cuestiones abiertas a su alrededor.

This paper introduces a subjective logic based argumentation framework primarily targeted at evidential reasoning. The framework explicitly caters for argument schemes, accrual of arguments, and burden of proof; these concepts appear in many types of argument, and are particularly useful in dialogues revolving around evidential reasoning. The concept of a sensor is also useful in this domain, representing a source of evidence, and is incorporated in our framework. We show how the framework copes with a number of problems that ...

The evidential reasoning (ER) rule for multi-attribute classification has been recently developed, and has enhanced the properties of the Dempster's rule through defining the weight and reliability of the evidence. This paper aims to introduce the ER rule for data classification and ensemble learning in the domain of trauma research. Moreover, it aims to identify multiple methods for building trauma prediction models, and to increase model accuracy in order to enhance the care services provided to trauma patients. The model proposed in this paper includes age, gender, age and gender combined, injury severity score, Glasgow coma scale, and modified Charlson comorbidity index as predictors of patient outcomes at 30 days or at discharge, whichever occurred first. The results of some machine learning (ML) algorithms such as decision tree, random forest, and artificial neural networks are compared to logistic regression results. The area under the curve (AUC) result of the artificial neural networks algorithm is 0.9076, which outperforms that of the logistic regression presented in the paper, i.e. 0.9045. Moreover, the application of the ER rule for ensemble learning indicates adequate prediction performance.

Witness credibility is important for establishing testimonial value. The story model posits that people construct narratives from evidence but does not explain how credibility is assessed. Formal approaches use Bayesian networks (BN) to represent legal evidence. Recent empirical work suggests people might also reason using qualitative causal networks. In two studies, participants read a realistic trial transcript and judge guilt and witness credibility. Study 1 varied testimonial consistency and defendant character. Guilt and credibility assessments were affected by consistency but not prior convictions. Study 2 constructed a BN to represent consistency issues. Individual parameter estimates were elicited for the corresponding BN to compute posterior predictions for guilt and credibility. The BN provided a good model for overall and individual guilt and credibility ratings. These results suggest people construct causal models of the evidence and consider witness credibility. The BN approach is a promising direction for future research in legal reasoning.

Recent research (Pilditch, Fenton, & Lagnado, 2019) shows that people are susceptible to zero-sum thinking in evidence evaluation, where they dismiss or underweight the probative value of evidence that is equally predicted by multiple independent hypotheses. But such an assumption is only valid when explanations are mutually exclusive and exhaustive. The present work extends these findings by looking at the context of information selection, and the decisional consequences of the zero-sum fallacy. It uses an information metric to quantify the cost of the error in terms of overlooked information.

The evidence theory and its propositional conflict redistribution variant rules are mathematical formalisms used to represent uncertain as well as ambiguous data. The evidence combination rules proposed in these formalisms do not satisfy the idempotence property. However, in a variety of applications, it is desirable that the evidence combination rules satisfy this property. In response to this challenge, the present work proposes a new formalism for reasoning under uncertainty based on new consensus and conflicting of evidence concepts. This mathematical formalism is evaluated using a real world activity recognition problem in smart home environment. The results show that one rule of our formalism respects the idempotence property and improves the accuracy of activity recognition.

With an aim of having a normal behavior in combination of bodies of evidence, this paper proposes a new combination rule which includes the cardinality of focal set elements in conjunctive operation and the conflict redistribution to all meaningful sets steps. This distribution is based on factors, which are computed by weighting the masses assigned to each set by the sources of information and those obtained by the conjunctive operation with its cardinality. This strategy forces the conflict redistribution in favor of the more committed hypothesis. Our method is evaluated and compared with some numerical examples reported in the literature. As result, this rule redistributes the conflict in favor of the more committed hypothesis and gives intuitive interpretation for combining multiple information sources with coherent results.

Abstract: In this paper, a clustering based machine learning algorithm called Clustering Algorithm System (CAS) is introduced. The CAS algorithm is tested to evaluate its performance and find fruitful results. We have been presented some heuristics to facilitate machine-learning ...

The location of any business facility plays a pivotal role in the success of the business and it is one of the major considerations in subjects such as business strategy, operations management and economics. This paper proposes the use of Multi-criteria Decision Analysis, as an effective means of tackling the crucial location decision for a manufacturing plant. An exploratory case of a strawberry syrup manufacturer is investigated and a ranking of three speculative locations where the company can establish their new strawberry syrup production plant is presented here. To evaluate the effectiveness of MCDA in determining the optimum set of locations for the plant, extensive sensitivity analyses were carried out with all the criteria involved, and this further supported the findings of the three most possible locations. Finally, the paper proposes the development of a risk continuum based on extreme risk preferences, which can be used as a tool to determine the ranking of the alternatives based on the risk preference and attitude. The paper showed the potential of the MCDA as an effective tool in location decisions, presents a tool to incorporate diversity in decision maker preferences, and then sets the stage for further research in this specific branch. JEL Classifications: C3, M2, D7 and D8

Machine-learning research is to study and apply the computer modeling of learning processes in their multiple manifestations, which facilitate the development of intelligent system. In this paper, we have introduced a clustering based machine-learning algorithm called ...

This paper describes an on-going effort to build an Adaptable Data Fusion Testbed (ADFT) based on a Knowledge-Based System (KBS) BlackBoard (BB) architecture to perform data fusion of imaging and nonimaging sensors present on-board the CP-140 Canadian maritime patrol aircraft. The algorithms incorporate state-of-the-art tracking in clutter and evidential reasoning for target identification. The end result offers the user a

Various demographic and medical factors can be linked to severe deterioration of patients suffering from traumatic injuries. Accurate identification of the most relevant variables is essential for building more accurate prediction models and making more rapid life-saving medical decision. The intention of this paper is to select a number of features that can be used to accurately predict patients’ outcomes through three feature selection methods: random forest, ReliefF and the evidential reasoning (ER) rule. The impact of an outcome’s class imbalance on feature selection is discussed, and synthetic minority over-sampling technique (SMOTE) is performed to show the differences in the selected features. The results show that length of stay in hospital, length of stay in intensive care unit, age, and Glasgow Coma Scale (GCS) are the most selected features across different techniques. The prediction models based on the features selected by the ER rule show the highest prediction performance represented by the area under the receiver operating characteristic curve (AUC) values, which has a median of 0.895 for the model employed by the ten highest-weighted variables, while the median AUC values are 0.827 and 0.885 if the ten highest-weighted variables are selected by ReliefF and random forest respectively. The results also show that after the ten most important features, increasing the number of the less important features has only a slight increase in prediction accuracy.

Most inferentialists hope to bypass givenness by tracking the conditionals claimants are implicitly committed to. I argue that this approach is underdetermined because one can always construct parallel trees of conditionals. I illustrate this using the Müller-Lyer illusion and touching a table. In the former case, the lines are either even or uneven; in the latter case, a moving hand will either sweep through or be halted. For each possibility, we can rationally foresee consequents. However, I argue that, until and unless we benefit from what is given in experience, we cannot know whether to affirm the antecedents of those conditionals.