In this paper, we focus on allying fuzzy logic, which is a suitable model for human-like information, and causality, which is a key concept for humans to generate knowledge from observations and to build explanations. If a fuzzy premise causes a fuzzy consequence, then acting on the fuzzy premise will have an impact on the fuzzy consequence. This is not necessarily the case for common fuzzy rules whose induction is based on correlation. Indeed, correlations may be due to some latent common cause of fuzzy premise and consequence. In this case, a change in the value of the fuzzy premise may not affect the fuzzy consequence as it should. We propose an approach to construct a set of causality-based fuzzy rules from crisp observational data. The idea is to identify causal relationships on the set of fuzzified inputs and outputs by well-known constraints-based causal discovery algorithms such as Peter-Clark and Fast Causal Inference. The causal discovery algorithms are combined with entropy-based conditional independent testing that avoids making hypotheses on the data distribution. Experiments are conducted to evaluate our approach in terms of ability to recover causal relationships between fuzzy sets in the presence of a latent common cause. The results illustrate the interest of our approach compared to a correlation-based approach and state-of-the-art approaches.

In this paper, we explore the min-max inference mechanism of any rule-based system of n if-then possibilistic rules. We establish an additive formula for the output possibility distribution obtained by the inference. From this result, we deduce the corresponding possibility and necessity measures. Moreover, we give necessary and sufficient conditions for the normalization of the output possibility distribution. As application of our results, we tackle the case of a cascade of two if-then possibilistic rules sets and establish an input-output relation between the two min-max equation systems. Finally, we associate to the cascade construction an explicit min-max neural network.

Fuzzy logic has been successfully used in various crisis management systems. In such systems, the geographical aspect is usually very important and relies on Geographical Information Systems. Most of the approaches are focused on 2D information. In this paper, we use the fuzzy morpho-mathematics framework to define new relations to reason on the topography with a digital terrain model. In particular, we focus on the characterisation of the line of greatest dip. Without loss of generality, we then illustrate those relations on a case of runoff from a building and a terrain.

Semantic image annotation is a field of paramount importance in which deep learning excels. However, some application domains, like security or medicine, may need an explanation of this annotation. Explainable Artificial Intelligence is an answer to this need. In this work, an explanation is a sentence in natural language that is dedicated to human users to provide them clues about the process that leads to the decision: the labels assignment to image parts. We focus on semantic image annotation with fuzzy logic that has proven to be a useful framework that captures both image segmentation imprecision and the vagueness of human spatial knowledge and vocabulary. In this paper, we present an algorithm for textual explanation generation of the semantic annotation of image regions.

In this paper, we present a new approach to predict monotonous functions based on approximate reasoning and in particular on the Gradual Generalized Modus Ponens (GGMP) in fuzzy logic. We propose to optimise the parameters of such fuzzy rules with a genetic algorithm considering few experimental data. We use our approach to predict some properties of materials from their manufacturing process parameters. We automatically extract causality, seek for graduality and then set up the GGMP. We tested on both toy and real world datasets. We also discuss the importance of gradual knowledge in materials science.

Decision support systems coupled with a Geographic Information System (GIS) are useful for assessing spatialized risks, which can be either natural, industrial or even anthropic. However, when the knowledge is not sufficiently precise or when the processes are complex, the methods of analysis of the GIS are not always sufficient. In this article, we propose to couple a GIS with a Fuzzy Expert System capable of evaluating fuzzy spatial rules, that is to say, fuzzy rules integrating spatial relations or properties. These are based on fuzzy mathematical morphology and model high-level metric or topological relationships such as “being close to”, “being in the direction of” or “being adjacent to”. The proposed approach has been applied to the identification of risky issues in the case of wildfires. We detail the expert knowledge modeled and show the results obtained in a graphic interface dedicated to the intervention of firefighters.

Chemical vapor analysis devices are booming, thanks to a growing need in areas such as security and quality control. These control devices are based on various technologies that are the subject of important researches in an ever-growing community of physicists and electronics. However, the data from these sensors are often processed by conventional algorithms poorly configured for the purpose of automatically recognizing target chemical compounds. These algorithms are often based on statistical models that are not always adapted to a limited number of learning data and demonstrated reproducibility problems for these kind of sensors. In this article, we propose to train fuzzy models and compare their performances with the classical methods of the state of the art, to show how practical they can be for such applications. Three different uses cases will be studied: toxic chemicals recognition, detection of counterfeit coffee capsules and detection of a chemical weapon among everyday products.

Fuzzy decision trees have been successfully applied in numerous domains. The popularity of these models comes notably from their interpretability, namely the ability of humans to understand them. However, on the contrary to neural networks, the induction of such models does not include a generation of their own feature space. In this work, the embedding of feature construction in fuzzy decision tree induction algorithms is studied, so that they can create new input features, without affecting the overall interpretability of the model. This method is successfully applied to a classification problem in high-energy physics to study the benefits of having constructed features in fuzzy decision tree on the classification scores, allowing them to have their own interpretable representation of the data.

A good feature representation is a determinant factor to achieve high performance for many machine learning algorithms in terms of classification. This is especially true for techniques that do not build complex internal representations of data (e.g. decision trees, in contrast to deep neural networks). To transform the feature space, feature construction techniques build new high-level features from the original ones. Among these techniques, Genetic Programming is a good candidate to provide interpretable features required for data analysis in high energy physics. Classically, original features or higher-level features based on physics first principles are used as inputs for training. However, physicists would benefit from an automatic and interpretable feature construction for the classification of particle collision events.Our main contribution consists in combining different aspects of Genetic Programming and applying them to feature construction for experimental physics. In particular, to be applicable to physics, dimensional consistency is enforced using grammars.Results of experiments on three physics datasets show that the constructed features can bring a significant gain to the classification accuracy. To the best of our knowledge, it is the first time a method is proposed for interpretable feature construction with units of measurement, and that experts in high-energy physics validate the overall approach as well as the interpretability of the built features.

As Artificial Intelligence and fuzzy systems are at the center of the emergence of advanced technologies such as autonomous vehicles or medical decision support systems, a problem of trust from a human point of view is strongly appearing. In this article, we tackle the problem of explanation of a fuzzy inference system decision in its entirety: from the conception of an algorithm that produces a textual explanation to its evaluation.We define a function which is able to associate to any activated fuzzy rule, the structure responsible of its activation degree. To assess our method, we defined a protocol to evaluate AIgenerated explanation, and made an experiment: explanations obtained from the classification of pastas. Despite limitations, the results show a good transparency of the reasoning, consistency and good global effectiveness in generated explanations.

Adapting User Interfaces to various technological devices is most often than not a part of human-computer interaction requirements. Although many studies addressed this topic some challenges remain, such as context uncertainty and combination of adaptation rules. This article represents an attempt at tackling these challenges, using fuzzy logic to handle adaptation. It proposes an architecture where an adaptation engine is supported by both fuzzy logic and Boolean logic, and illustrated by a prototype. The relevance of such approach has been studied through a theoretical comparison and an experiment including eight experts.

The goal of explainable artificial intelligence is to solve problems in a way that humans can understand how it does it. However, few approaches have been proposed so far and some of them lay more emphasis on interpretability than on explainability. In this paper, we propose an approach that is based on learning fuzzy relations and fuzzy properties. We extract frequent relations from a dataset to generate an explained decision. Our approach can deal with different problems, such as classification or annotation. A model was built to perform explained classification on a toy dataset that we generated. It managed to correctly classify examples while providing convincing explanations. A few areas for improvement have been spotted, such as the need to filter relations and properties before or while learning them in order to avoid useless computations.

The democratization of localization sensors allows to get the coordinates of mobile entities at a cheap cost but with a variable precision. Mobile devices are almost all equipped with such sensors and it is now possible to locate humans, animals or vehicles in order to monitor their activities.In this paper, we present fuzzy online relations which allow to describe some behaviors of a mobile entity from its coordinates and regarding a spatial region described as a crisp or a fuzzy geometry. Those operators are used within a fuzzy expert system in order to make decisions like yielding alerts, summarizing activities or assessing performance measures, regarding the behavior of localized entities.

In this paper we propose a method of explainable material classification from imprecise chemical compositions. The problem of classification from imprecise data is addressed with a fuzzy decision tree whose terms are learned by a clustering algorithm. We deduce fuzzy rules from the tree, which will provide a justification of the result of the classification. Two opposed approaches are compared : the probabilistic approach and the possibilistic approach.

Online fuzzy expert systems can be used to process data and event streams, providing a powerful way to handle their uncertainties and their inaccuracy. Moreover, human experts can decide how to process the streams with rules close to natural language. However, to extract high level information from these streams, they need at least to describe the temporal relations between the data or the events. In this paper, we propose a straightforward way to design temporal operators which relies on the mathematical definition of some base operators and then their combination into more sophisticated operators to assess precedence, periodicity or persistence. We also introduce the concept of expiration of temporal expressions on online fuzzy expert systems, that is to say the capacity to change the values of outputs whereas the inputs have not changed.

In recent years, artificial intelligence tools have democratized and are more and more often used by people who are not experts in the field. For instance, systems based on rules or constraints require human expertise as input to replicate the desired behavior. Despite the explosion of new devices and new input paradigms, such as tablets and other touch interfaces, usability of these tools seems not to have taken advantage of these recent advances. In this article, we focus on a fuzzy expert system for which users want to enter rules. We use our industrial partnerships to define with current users their needs in terms of rule authoring. They expressed their will of more mobility, more modernism, less mathematics. We present our work that involves the use of new touch interfaces to capture a fuzzy rule with only one finger. We end this article by the evaluation of the GUI with a user panel.

We propose in this paper an original approach to the TV stream structuring problem. The goal of our work is to automatically break the TV stream into telecasts and advertisings and to label each telecast with its genre. One can think the TV stream structuring problem can be solved by an alignment of the program guide on the stream. But our study shows that, in average, only 25% of the telecasts per day are presented in the program guide. Hence, our method consists in improving statistically these program guides in order to reduce the TV stream structuring problem to a simple alignment problem. The improvement consists in adding the missing telecasts. We present an original system that lays on the modeling of past TV schedules by a contextual hidden Markov model and a regression tree. Interesting results are presented at the end of the paper

The French National Audiovisual Institute is in charge of archiving continuously the video stream of every French channel. In order to be described, each program must be isolated in the stream. Our paper focuses on creating a system which can find programs' boundaries. We propose in this article a way to predict various programs' boundaries to obtain temporal windows in which our system is able to search