Amel Borgi

Extended symbolic approximate reasoning based on linguistic modifiers

Approximate reasoning allows inferring with imperfect knowledge. It is based on a generalization of modus ponens (MP) known as generalized modus ponens (GMP). We are interested in approximate reasoning within symbolic multi-valued logic framework. In a previous work, we have proposed a new GMP based on linguistic modifiers in the multi-valued logic framework. The use of linguistic modifiers allows having a gradual reasoning; moreover, it allows checking axiomatics of approximate reasoning. In this paper, we extend our approximate reasoning to hold with complex rules, i.e., rules whose premises are conjunction or disjunction of propositions. For this purpose, we introduce a new operator that aggregates linguistic modifiers and verifies the required properties of logical connectives within the multi-valued logic framework.

FSCEP: A New Model for Context Perception in Smart Homes

With the emergence of the Internet of Things and smart devices, smart homes are becoming more and more popular. The main goal of this study is to implement an event driven system in a smart home and to extract meaningful information from the raw data collected by the deployed sensors using Complex Event Processing (CEP). These high-level events can then be used by multiple smart home applications in particular situation identification. However, in real life scenarios, low-level events are generally uncertain. In fact, an event may be outdated, inaccurate, imprecise or in contradiction with another one. This can lead to misinterpretation from CEP and the associated applications. To overcome these weaknesses, in this paper, we propose a Fuzzy Semantic Complex Event Processing (FSCEP) model which can represent and reason with events by including domain knowledge and integrating fuzzy logic. It handles multiple dimensions of uncertainty, namely freshness, accuracy, precision and contradiction. FSCEP has been implemented and compared with a well known CEP. The results show how some ambiguities are solved.

Camphor odor recognition within unbalanced multi-sets

In fuzzy logic context, some works deal with the camphor odor perception. In this paper, we present a novel rule-based decision system for the camphor odor recognition within unbalanced multiset. Our first contribution consists in an adaptation of fuzzy knowledge representation and inference rules to the multi-valued logic context. The second contribution concerns the improvement of the knowledge base by changing facts representation and adding new rules. This proposition provides satisfactory results in term of precision, recall, F-measure and accuracy.

A fuzzy semantic CEP model for situation identification in smart homes

Uncertainty is an essential issue for smart home applications. Events generated from sensors can be outdated, inaccurate, imprecise or in contradiction with other ones. These unreliable data can lead to dysfunction in smart home applications. To tackle these challenges, we propose a new model named FSCEP (Fuzzy Semantic Complex Event Processing) that integrates fuzzy logic paradigm, semantic features through an ontology and traditional CEP. We confronted FSCEP with other works tackling uncertainty for CEP and experimented it through simulation with early but promising result

SimNCD: An agent-based formalism for the study of noncommunicable diseases

Abstract Noncommunicable diseases (NCDs) are multifactorial chronic illnesses that cause long-term morbidities in a persons life. Their nonlinear and complex dynamics make them difficult to control and predict. Hence, it is crucial for public health researchers and practitioners to understand and model them. This paper proposes a generic interaction-oriented agent-based modeling for NCDs, called SimNCD. It models individuals living within a social network and daily engaging in activities from the physical environment. The practice of these activities stirs the individuals health vulnerability to cure, acquire or maintain NCDs. This modeling can serve as a tool in public health policies for the study of NCDs where the individuals behavior greatly influences the factors predisposing to that disease. We also propose a specified version of SimNCD for modeling childhood obesity, called SimNCDChO. More precisely, the latter models the complex relationships between childrens physical activity and the development of their corpulence during their growth.

Representation of unbalanced terms in multi-valued logic

Various approaches were proposed to represent and treat imperfect knowledge, in particular fuzzy logic and multivalued logic. Such knowledge is generally expressed with uniformly distributed linguistic term sets. However, in many cases, we need to describe information with unbalanced term sets. In our work we introduce a new approach to represent such term sets. It corresponds to an algorithm which unifies data expressed in different multi-sets on a same uniform scale. This latter can be integrated into a linguistic reasoning process with unbalanced multi-sets (using linguistic modifiers, approximate reasoning,...).

Features' Associations in Fuzzy Ensemble Classifiers.

In this work, we are interested in ensemble methods for fuzzy rule-based classification systems where the decisions of different classifiers are combined to form the final classification model. We focus on ensemble methods that cluster the set of attributes into subgroups and treat each subgroup separately. This allows decomposing the learning problem into sub-problems of lower complexity and obtaining more intelligible rules as their number and size are smaller. In this paper, we study different methods that allow finding associations between the attributes. In this context, SIFRA is an interesting attributes regrouping method based on association rules concept. We compare SIFRA with some other association methods and show, via a detailed analysis of experimental results, that it is able to find interesting types of associations including linear and non-linear ones. Moreover, it improves the system’s accuracy and guarantees a smaller rules number compared to classical FRBCS.

Supervised methods for regrouping attributes in fuzzy rule-based classification systems

This paper focuses on ensemble methods for Fuzzy Rule-Based Classification Systems (FRBCS) where the decisions of different classifiers are combined in order to form the final classification model. The proposed methods reduce the FRBCS complexity and the generated rules number. We are interested in particular in ensemble methods which cluster the attributes into subgroups of attributes and treat each subgroup separately. Our work is an extension of a previous ensemble method called SIFRA. This method uses frequent itemsets mining concept in order to deduce the groups of related attributes by analyzing their simultaneous appearances in the databases. The drawback of this method is that it forms the groups of attributes by searching for dependencies between the attributes independently from the class information. Besides, since we deal with supervised learning problems, it would be very interesting to consider the class attribute when forming the attributes subgroups. In this paper, we proposed two new supervised attributes regrouping methods which take into account not only the dependencies between the attributes but also the information about the class labels. The results obtained with various benchmark datasets show a good accuracy of the built classification model.

Feature selection based on Choquet integral for human activity recognition

Human activity recognition (HAR) is an important research issue for pervasive computing that aims to identify human activities. Extracted features from raw sensors are often large and some of them can be irrelevant and redundant. Therefore, its important to perform feature selection to select the most relevant features in order to increase the recognition accuracy. However, classical feature selection methods are generally linear and sequential and they do not consider existing dependencies and interactions among activities (classes). To overcome this shortcoming, a feature selection based on Choquet integral for HAR is proposed in this paper. The Choquet integral is a non linear and a non additive method. Its employed to determine scores for features by modeling interactions between activities through the fuzzy measure theory. Classification results on HAR dataset using Random Forest classifier indicate that the recognition accuracy remains stable using half of the features. Moreover, classification performance is further improved.

Attributes regrouping by association rules in SUCRAGE

We are interested in a supervised learning method by automatic generation of classification rules: SUCRAGE. Premises construction is done by grouping the dependant attributes. This selection in one block of the features is realized by linear correlation research among the training set elements. Only numerical features can be taken into account by linear correlation search. In this article, we propose to extend SUCRAGE to handle symbolic attributes by using another method of regrouping attributes based on Association Rules. This method can detect different types of association between quantitative as well as qualitative attributes. The obtained results in generalization with various data using the built rules are very satisfactory.