Salsabil Trabelsi

Pruning belief decision tree methods in averaging and conjunctive approaches

The belief decision tree (BDT) approach is a decision tree in an uncertain environment where the uncertainty is represented through the Transferable Belief Model (TBM), one interpretation of the belief function theory. The uncertainty can appear either in the actual class of training objects or attribute values of objects to classify. From the procedures of building BDT, we mention the averaging and the conjunctive approaches. In this paper, we develop pruning methods of belief decision trees induced within averaging and conjunctive approaches where the objective is to cope with the problem of overfitting the data in BDT in order to improve its comprehension and to increase its quality of the classification.

Classification systems based on rough sets under the belief function framework

In this paper, we present two classification approaches based on Rough Sets (RS) that are able to learn decision rules from uncertain data. We assume that the uncertainty exists only in the decision attribute values of the Decision Table (DT) and is represented by the belief functions. The first technique, named Belief Rough Set Classifier (BRSC), is based only on the basic concepts of the Rough Sets (RS). The second, called Belief Rough Set Classifier, is more sophisticated. It is based on Generalization Distribution Table (BRSC-GDT), which is a hybridization of the Generalization Distribution Table and the Rough Sets (GDT-RS). The two classifiers aim at simplifying the Uncertain Decision Table (UDT) in order to generate significant decision rules for classification process. Furthermore, to improve the time complexity of the construction procedure of the two classifiers, we apply a heuristic method of attribute selection based on rough sets. To evaluate the performance of each classification approach, we carry experiments on a number of standard real-world databases by artificially introducing uncertainty in the decision attribute values. In addition, we test our classifiers on a naturally uncertain web usage database. We compare our belief rough set classifiers with traditional classification methods only for the certain case. Besides, we compare the results relative to the uncertain case with those given by another similar classifier, called the Belief Decision Tree (BDT), which also deals with uncertain decision attribute values.

Heuristic method for attribute selection from partially uncertain data using rough sets

In this paper, we deal with the problem of attribute selection from partially uncertain data based on rough sets without costly calculation. The uncertainty exists in decision attributes and is represented by the transferable belief model, one interpretation of the belief function theory. To solve this problem, we propose a heuristic method for attribute selection able to extract the more relevant features needed in the classification process. The simplification of the uncertain decision table using this heuristic method yields to learn simplified and more significant belief decision rules in a quick time. The experiments show interesting results based on two evaluation criteria such as the accuracy classification and the time complexity.

Belief Rough Set Classifier

In this paper, we propose a new rough set classifier induced from partially uncertain decision system. The proposed classifier aims at simplifying the uncertain decision system and generating more significant belief decision rules for classification process. The uncertainty is reperesented by the belief functions and exists only in the decision attribute and not in condition attribute values.

Dynamic Reduct from Partially Uncertain Data Using Rough Sets

In this paper, we deal with the problem of attribute selection from a sample of partially uncertain data. The uncertainty exists in decision attributes and is represented by the Transferable Belief Model (TBM), one interpretation of the belief function theory. To solve this problem, we propose dynamic reduct for attribute selection to extract more relevant and stable features for classification. The reduction of the uncertain decision table using this approach yields simplified and more significant belief decision rules for unseen objects.

Rule discovery process based on rough sets under the belief function framework

In this paper, we deal with the problem of rule discovery process based on rough sets from partially uncertain data. The uncertainty exists only in decision attribute values and is handled by the Transferable Belief Model (TBM), one interpretation of the belief function theory. To solve this problem, we propose in this uncertain environment, a new method based on a soft hybrid induction system for discovering classification rules called GDT-RS which is a hybridization of the Generalization Distribution Table and the Rough Set methodology.

Heuristic for attribute selection using belief discernibility matrix

This paper proposes a new heuristic attribute selection method based on rough sets to remove the superfluous attributes from partially uncertain data. We handle uncertainty only in decision attributes (classes) under the belief function framework. The simplification of the uncertain decision table which is based on belief discernibility matrix generates more significant attributes with fewer computations without making significant sacrifices in classification accuracy.

A comparison of dynamic and static belief rough set classifier

In this paper, we propose a new approach of classification based on rough sets denoted Dynamic Belief Rough Set Classifier (D-BRSC) which is able to learn decision rules from uncertain data. The uncertainty appears only in decision attributes and is handled by the Transferable Belief Model (TBM), one interpretation of the belief function theory. The feature selection step of the construction procedure of our new technique of classification is based on the calculation of dynamic reduct. The reduction of uncertain and noisy decision table using dynamic approach which extracts more relevant and stable features yields more significant decision rules for the classification of the unseen objects. To prove that, we carry experimentations on real databases using the classification accuracy criterion. We also compare the results of D-BRSC with those obtained from Static Belief Rough Set Classifier (S-BRSC).

Exhaustive Search with Belief Discernibility Matrix and Function

This paper proposes a new feature selection method based on rough sets to take away the unnecessary attributes for the classification process from partially uncertain decision system. The uncertainty exists only in the decision attributes (classes) and is represented by the belief function theory. The simplification of the uncertain decision table to generate more significant attributes is based on computing all possible reducts. To obtain these reducts, we propose a new definition of the concepts of discernibility matrix and function under the belief function framework. Experimentations have been done to evaluate this exhaustive solution.

Classification with dynamic reducts and belief functions

In this paper, we propose two approaches of classification namely, Dynamic Belief Rough Set Classifier (D-BRSC) and Dynamic Belief Rough Set Classifier based on Generalization Distribution Table (D-BRSC-GDT). Both the classifiers are induced from uncertain data to generate classification rules. The uncertainty appears only in decision attribute values and is handled by the Transferable Belief Model (TBM), one interpretation of the belief function theory. D-BRSC only uses the basic concepts of Rough Sets (RS). However, D-BRSC-GDT is based on GDT-RS which is a hybridization of Generalization Distribution Table (GDT) and Rough Sets (RS). The feature selection step relative to the construction of the two classifiers uses the approach of dynamic reduct which extracts more relevant and stable features. The reduction of uncertain and noisy decision table using dynamic approach generates more significant decision rules for the classification of unseen objects. To prove that, we carry experimentations on real databases according to three evaluation criteria including the classification accuracy. We also compare the results of D-BRSC and D-BRSC-GDT with those obtained from Static Belief Rough Set Classifier (S-BRSC) and Static Belief Rough Set Classifier based on Generalization Distribution Table (S-BRSC-GDT). To further evaluate our rough sets based classification systems, we compare our results with those obtained from the Belief Decision Tree (BDT).