Sandra A. Sandri

Elicitation, assessment, and pooling of expert judgments using possibility theory

The problem of modeling expert knowledge about numerical parameters in the field of reliability is reconsidered in the framework of possibility theory. Usually expert opinions about quantities such as failure rates are modeled, assessed, and pooled in the setting of probability theory. This approach does not seem to always be natural since probabilistic information looks too rich to be currently supplied by individuals. Indeed, information supplied by individuals is often incomplete, imprecise rather than tainted with randomness. Moreover, the probabilistic framework looks somewhat restrictive to express the variety of possible pooling modes. In this paper, the authors formulate a model of expert opinion by means of possibility distributions that are thought to better reflect the imprecision pervading expert judgments. They are weak substitutes to unreachable subjective probabilities. Assessment evaluation is carried out in terms of calibration and level of precision, respectively, measured by membership grades and fuzzy cardinality indexes. Finally, drawing from previous works on data fusion using possibility theory, the authors present various pooling modes with their formal model under various assumptions concerning the experts. A comparative experiment between two computerized systems for expert opinion analysis has been carried out, and its results are presented in this paper. >

Fuzzy Temporal/Categorical Information in Diagnosis

This paper proposes a way of incorporating fuzzy temporal reasoning within diagnostic reasoning. Disorders are described as an evolving set of necessary and possible manifestations. Ill-known moments in time, e.g., when a manifestation should start or end, are modeled by fuzzy intervals, which are also used to model the elapsed time between events, e.g., the beginning of a manifestation and its end. Patient information about the intensity and times in which manifestations started and ended are also modeled using fuzzy sets. The paper discusses many measures of consistency between the patient/s data and the disorder model, and defines when the manifestations of the patient can be explained by a disorder. This work also discusses related issues such as the intensity of manifestations and the speed in which the disorder is evolving, given the patients data, and how to use that information to make predictions about future and past events.

Specification, Mapping and Control for QoS Adaptation

In this paper we describe a fuzzy-control approach for quality of service (QoS) adaptation, needed in distributed multimedia applications. QoS adaptation is necessary (a) due to sudden variations in network resource availability, especially in the case of Internet, and (b) due to multiple applications requiring shared resource such as bandwidth. To solve the problem of QoS adaptation, several sub-problems need to be considered: (1) mapping of user perception and different combinations of application QoS values onto a uniform quality metric, (2) estimation, control and adjustment of application QoS parameters in case of network and other resource congestion, and (3) enforcement algorithm which reacts according to adapted QoS parameters. Our approach is to solve the QoS adaptation using the integration of (a) quality degree function, which maps the application QoS parameters into a metric, called quality degree, (b) fuzzy controller, which controls, estimates and adjusts the application QoS parameters according to resource availability, and (c) filter algorithms, which are the services to enforce the adapted QoS parameters. The quality degree function associates quality degree as the quality measure with different combinations of application QoS values. This function is influenced by the user’s perception of quality. The fuzzy control takes the results of the quality degree function, estimates the new quality degree and its corresponding quality level, predicts the new application QoS parameters, and adjusts them. The results of the adapted QoS parameters are then used by the filter algorithms to enforce the changes, proposed by the fuzzy controller, by allocating bandwidth to the application according to its QoS parameter values. We have implemented and applied the quality degree function, the fuzzy controller, and the filter algorithms to the video distribution system (VDS). The results of VDS over the local area network show that (1) the target system improves user perceived QoS at the receivers, and (2) the bandwidth utilization increases significantly when using our fuzzy-control approach for QoS adaptation.

Fuzzy ant colony optimization for estimating chlorophyll concentration profile in offshore sea water

The determination of some inherent optical properties can be addressed by estimating the ocean chlorophyll concentration, if bio-optical models can be applied – such as for the offshore sea water. This inverse problem can be formulated as an optimization problem and iteratively solved, where the radiative transfer equation is the direct model. An objective function is given by the square difference between computed and in situ experimental radiances at every iteration. In the standard ant colony optimization (ACO), the pheromone is reinforced only on the best ant of the population. The fuzzy strategy consists in including additional pheromone quantity on the best ant, but a small pheromone quantity is also spread over the other solutions close to the best one. Test results show that the fuzzy-ACO produces better inverse solutions.

Risk Mapping of Schistosomiasis in Minas Gerais, Brazil, Using MODIS and Socioeconomic Spatial Data

Schistosomiasis mansoni is a disease with social and behavioral characteristics. Snails of the Biomphalaria species, the diseases intermediate host, use water as a vehicle to infect man, the disease main host. In Brazil, six million people are infected. From 1995 to 2005, more than a million positive cases were reported, 27% of them in the state of Minas Gerais. The objective of this paper is to estimate the prevalence risk of schistosomiasis, in terms of remote sensing, climate, socioeconomic, or neighborhood variables or a subset of them. We present two approaches for modeling and classifying the infection risk: a global and a regional one, both of them using the aforementioned variables. In the first approach, a unique regression model was generated and used to estimate the disease risk for the entire state. In the second approach, the state was divided in four regions, and a model was generated for each of them. The first model obtained 47.2% of overall accuracy (AC) and the second achieved 62.4%, which were considered unsatisfactory. To improve these results, the concept of imprecise classification, defined in terms of the standard deviation of estimates and several reliability levels, is used for the generation of two imprecise classification maps. The AC for the imprecise classification was 83.8% for the global model and 91.9% for the regional one, which were now considered acceptable. Particularly, regionalization has proven to be a good guideline to follow in future works involving geographical aspects and large data heterogeneity.

Fuzzy expert systems architecture for image classification using mathematical morphology operators

We propose a fuzzy expert systems architecture for image classification, whose rules are implemented through translation invariant mathematical morphology operators. The use of the architecture is illustrated by an expert system that classifies an area of the Tapajos National Forest, in Brazil.

Restoring Consistency in Systems of Fuzzy Gradual Rules Using Similarity Relations

We present here a method that uses similarity relations to restore consistency in fuzzy gradual rules systems: we propose to transform potentially inconsistent rules by making their consequents more imprecise. Using a suitable similarity relation we obtain consistent rules with a minimum of extra imprecision. We also present an application to illustrate the approach.

An annotated logic theorem prover for an extended possibilistic logic

In this paper we present a theorem prover for possibilistic logic extended with fuzzy constants and fuzzily restricted quantifiers. First of all, we propose a representation for this logic in terms of Horn clauses. We then show how to transform this Horn clause formalism into the generalized annotated logic formalism proposed by Kifer and Subrahmanian in 1992. Specifically, the valuation in a possibilistic clause generates the annotation which will be attached to the head of the annotated clause. We also show how the inference rules of this possibilistic logic can be translated in terms of the mechanisms provided by this annotated logic. Finally, we discuss the implementation of a theorem prover for this possibilistic logic, now translated to annotated logic, in system KOMET, a large-purpose system which has generalized annotated logic as its underlying framework. In this paper we also present proofs relative to the theoretical issues and some examples implemented in KOMET.

Tunable equivalence fuzzy associative memories

This paper introduces a new class of fuzzy associative memories (FAMs) called tunable equivalence fuzzy associative memories, for short tunable E-FAMs or TE-FAMs, that are determined by the application of parametrized equivalence measures in the hidden nodes. Tunable E-FAMs belong to the class of ?-FAMs that have recently appeared in the literature. In contrast to previous ?-FAM models, tunable E-FAMs allow for the extraction of a fundamental memory set from the training data by means of an algorithm that depends on the evaluation of equivalence measures. Furthermore, we are able to optimize not only the weights corresponding to the contributions of the hidden nodes but also the contributions of the attributes of the data by tuning the parametrized equivalence measures used in a TE-FAM model. The computational effort involved in training tunable TE-FAMs is very low compared to the one of the previous ?-FAM training algorithm.

A method for deriving order compatible fuzzy relations from convex fuzzy partitions

We address a special kind of fuzzy relations capable of modeling that two elements in the universe of discourse are similar to the extent that they are close to each other with respect to a given total order. These order compatible fuzzy relations are reflexive and symmetric but not necessarily T-transitive. We address the requirements to construct such relations from a large class of fuzzy partitions that obey some requirements from the fuzzy sets in the partition, such as convexity, that are useful but not severely constraining. We also propose a new method to obtain those fuzzy relations from these partitions.