Institut National des Sciences Appliquées de Toulouse

In an industrial maintenance context, degradation diagnosis is the problem of determining the current level of degradation of operating machines based on measurements. With the emergence of Machine Learning techniques, such a problem can now be solved by training a degradation model offline and by using it online. While such models are more and more accurate and performant, they are often black-box and their decisions are therefore not interpretable for human maintenance operators. On the contrary, interpretable ML models are able to provide explanations for the model’s decisions and consequently improves the confidence of the human operator about the maintenance decision based on these models. This paper proposes a new method to quantitatively measure the interpretability of such models that is agnostic (no assumption about the class of models) and that is applied on degradation models. The proposed method requires that the decision maker sets up some high level parameters in order to measure the interpretability of the models and then can decide whether the obtained models are satisfactory or not. The method is formally defined and is fully illustrated on a decision tree degradation model and a model trained with a recent neural network architecture called Multiclass Neural Additive Model.

On considère l’opérateur elliptique <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="block"> <mml:mrow> <mml:mi>L</mml:mi> <mml:mi>u</mml:mi> <mml:mo>=</mml:mo> <mml:mo>-</mml:mo> <mml:mo>(</mml:mo> <mml:msub> <mml:mi>a</mml:mi> <mml:mrow> <mml:mi>i</mml:mi> <mml:mi>j</mml:mi> </mml:mrow> </mml:msub> <mml:msub> <mml:mi>u</mml:mi> <mml:msub> <mml:mi>x</mml:mi> <mml:mi>i</mml:mi> </mml:msub> </mml:msub> <mml:mo>+</mml:mo> <mml:msub> <mml:mi>d</mml:mi> <mml:mi>j</mml:mi> </mml:msub> <mml:mi>u</mml:mi> <mml:msub> <mml:mo>)</mml:mo> <mml:msub> <mml:mi>x</mml:mi> <mml:mi>j</mml:mi> </mml:msub> </mml:msub> <mml:mo>+</mml:mo> <mml:mo>(</mml:mo> <mml:msub> <mml:mi>b</mml:mi> <mml:mi>i</mml:mi> </mml:msub> <mml:msub> <mml:mi>u</mml:mi> <mml:msub> <mml:mi>x</mml:mi> <mml:mi>i</mml:mi> </mml:msub> </mml:msub> <mml:mo>+</mml:mo> <mml:mi>c</mml:mi> <mml:mi>u</mml:mi> <mml:mo>)</mml:mo> </mml:mrow> </mml:math> où les coefficients sont des fonctions mesurables appartenant à des espaces <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:msup> <mml:mi>L</mml:mi> <mml:mo>*</mml:mo> </mml:msup> <mml:mrow> <mml:mo>(</mml:mo> <mml:mi>Ω</mml:mi> <mml:mo>)</mml:mo> </mml:mrow> </mml:mrow> </mml:math> convenables dans un ouvert borné <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>Ω</mml:mi> </mml:math> de <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msup> <mml:mrow> <mml:mi mathvariant="bold">R</mml:mi> </mml:mrow> <mml:mi>n</mml:mi> </mml:msup> </mml:math> . Le but principal est d’étendre un résultat [par W. Littman, G. Stampacchia et H. Weinberger] sur les points réguliers pour le problème de Dirichlet à des équations plus générales (§10). Le paragraphe 3 contient aussi un principe de maximum pour les solutions faibles. Le paragraphe 4 contient des majorations a priori dans <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:msup> <mml:mi>L</mml:mi> <mml:mi>p</mml:mi> </mml:msup> <mml:mrow> <mml:mo>(</mml:mo> <mml:mi>Ω</mml:mi> <mml:mo>)</mml:mo> </mml:mrow> </mml:mrow> </mml:math> des solutions. Les paragraphes 5 et 7 sont consacrés à l’extension du théorème de Giorgi sur la continuité höldérienne des solutions ; l’extension au bord est aussi envisagée. La généralisation d’un théorème de Moser sur l’inégalité de Harnack est considérée dans le paragraphe 8. Le paragraphe 9 contient l’étude de la fonction de Green.

Hybrid systems, which combine both continuous and discrete behavior, are used in many fields, including robotics, biological systems, and control systems. However, due to their complexity, finding an accurate model is a challenge. This paper discusses the usage of symbolic regression to learn hybrid systems from data and specifically analyses learning parameters for a recent algorithm. Symbolic regression is a powerful tool that can automatically discover accurate and interpretable mathematical models in the form of symbolic expressions. Models generated by symbolic regression are a valuable tool for system identification and diagnosis, e.g., to predict future system behavior or detect anomalies. A major opportunity of our approach is the ability to detect transitions between different continuous behaviors of a system directly based on the dynamics. From a diagnosis perspective, this can advantageously be used to detect the system entering fault modes and identify their models. This paper presents a parameter study for a symbolic regression based identification algorithm.

Production of biogas from different organic materials is a most interesting source of renewable energy. The biomethane potential (BMP) of these materials has to be determined to get insight in design parameters for anaerobic digesters. Although several norms and guidelines for BMP tests exist, inter-laboratory tests regularly show high variability of BMPs for the same substrate. A workshop was held in June 2015, in Leysin, Switzerland, with over 40 attendees from 30 laboratories around the world, to agree on common solutions to the conundrum of inconsistent BMP test results. This paper presents the consensus of the intense roundtable discussions and cross-comparison of methodologies used in respective laboratories. Compulsory elements for the validation of BMP results were defined. They include the minimal number of replicates, the request to carry out blank and positive control assays, a criterion for the test duration, details on BMP calculation, and last but not least criteria for rejection of the BMP tests. Finally, recommendations on items that strongly influence the outcome of BMP tests such as inoculum characteristics, substrate preparation, test setup, and data analysis are presented to increase the probability of obtaining validated and reproducible results.

The McKibben artificial muscle is a pneumatic device characterized by its high level of functional analogy with human skeletal muscle. While maintaining a globally cylindrical shape, the McKibben muscle produces a contraction force decreasing with its contraction ratio, as does skeletal muscle. The maximum force-to-weight ratio can be surprisingly high for a limited radial dimension and for a conventional pressure range. A 50 g McKibben muscle can easily develop more than 1000 N under 5 bar pressure for an external radius varying from about 1.5 to 3 cm. Thus, robotics specialists are interested in this well-adapted artificial muscle for motorizing powerful yet compact robot arms. The basic McKibben muscle static modeling developed in the paper, which is based on the three main parameters (i.e., initial braid angle, initial muscle length, and initial muscle radius) and includes a three-parameter friction model of the thread against itself, has shown its efficiency in both isometric and isotonic contraction.

Escherichia coli is a normal inhabitant of the human gut. However, E. coli strains of phylogenetic group B2 harbor a genomic island called "pks" that codes for the production of a polyketide-peptide genotoxin, Colibactin. Here we report that in vivo infection with E. coli harboring the pks island, but not with a pks isogenic mutant, induced the formation of phosphorylated H2AX foci in mouse enterocytes. We show that a single, short exposure of cultured mammalian epithelial cells to live pks(+) E. coli at low infectious doses induced a transient DNA damage response followed by cell division with signs of incomplete DNA repair, leading to anaphase bridges and chromosome aberrations. Micronuclei, aneuploidy, ring chromosomes, and anaphase bridges persisted in dividing cells up to 21 d after infection, indicating occurrence of breakage-fusion-bridge cycles and chromosomal instability. Exposed cells exhibited a significant increase in gene mutation frequency and anchorage-independent colony formation, demonstrating the infection mutagenic and transforming potential. Therefore, colon colonization with these E. coli strains harboring the pks island could contribute to the development of sporadic colorectal cancer.

The association of arbuscular mycorrhizal (AM) fungi with plant roots is the oldest and ecologically most important symbiotic relationship between higher plants and microorganisms, yet the mechanism by which these fungi detect the presence of a plant host is poorly understood. Previous studies have shown that roots secrete a branching factor (BF) that strongly stimulates branching of hyphae during germination of the spores of AM fungi. In the BF of Lotus, a strigolactone was found to be the active molecule. Strigolactones are known as germination stimulants of the parasitic plants Striga and Orobanche. In this paper, we show that the BF of a monocotyledonous plant, Sorghum, also contains a strigolactone. Strigolactones strongly and rapidly stimulated cell proliferation of the AM fungus Gigaspora rosea at concentrations as low as 10(-13) M. This effect was not found with other sesquiterperne lactones known as germination stimulants of parasitic weeds. Within 1 h of treatment, the density of mitochondria in the fungal cells increased, and their shape and movement changed dramatically. Strigolactones stimulated spore germination of two other phylogenetically distant AM fungi, Glomus intraradices and Gl. claroideum. This was also associated with a rapid increase of mitochondrial density and respiration as shown with Gl. intraradices. We conclude that strigolactones are important rhizospheric plant signals involved in stimulating both the pre-symbiotic growth of AM fungi and the germination of parasitic plants.

The genome sequence of the solvent-producing bacterium Clostridium acetobutylicum ATCC 824 has been determined by the shotgun approach. The genome consists of a 3.94-Mb chromosome and a 192-kb megaplasmid that contains the majority of genes responsible for solvent production. Comparison of C. acetobutylicum to Bacillus subtilis reveals significant local conservation of gene order, which has not been seen in comparisons of other genomes with similar, or, in some cases closer, phylogenetic proximity. This conservation allows the prediction of many previously undetected operons in both bacteria. However, the C. acetobutylicum genome also contains a significant number of predicted operons that are shared with distantly related bacteria and archaea but not with B. subtilis. Phylogenetic analysis is compatible with the dissemination of such operons by horizontal transfer. The enzymes of the solventogenesis pathway and of the cellulosome of C. acetobutylicum comprise a new set of metabolic capacities not previously represented in the collection of complete genomes. These enzymes show a complex pattern of evolutionary affinities, emphasizing the role of lateral gene exchange in the evolution of the unique metabolic profile of the bacterium. Many of the sporulation genes identified in B. subtilis are missing in C. acetobutylicum, which suggests major differences in the sporulation process. Thus, comparative analysis reveals both significant conservation of the genome organization and pronounced differences in many systems that reflect unique adaptive strategies of the two gram-positive bacteria.

The stochastic multi-armed bandit model is a simple abstraction that has proven useful in many different contexts in statistics and machine learning. Whereas the achievable limit in terms of regret minimization is now well known, our aim is to contribute to a better understanding of the performance in terms of identifying the m best arms. We introduce generic notions of complexity for the two dominant frameworks considered in the literature: fixed-budget and fixed-confidence settings. In the fixed-confidence setting, we provide the first known distribution-dependent lower bound on the complexity that involves information-theoretic quantities and holds when m is larger than 1 under general assumptions. In the specific case of two armed-bandits, we derive refined lower bounds in both the fixed-confidence and fixed-budget settings, along with matching algorithms for Gaussian and Bernoulli bandit models. These results show in particular that the complexity of the fixed-budget setting may be smaller than the complexity of the fixed-confidence setting, contradicting the familiar behavior observed when testing fully specified alternatives. In addition, we also provide improved sequential stopping rules that have guaranteed error probabilities and shorter average running times. The proofs rely on two technical results that are of independent interest : a deviation lemma for self-normalized sums (Lemma 19) and a novel change of measure inequality for bandit models (Lemma 1).

The support plays an important role for supported metal catalysts by positioning itself as a macromolecular ligand, which conditions the nature of the active site and contributes indirectly but also sometimes directly to the reactivity. Metal species such as nanoparticles, clusters, or single atoms can be deposited on carbon materials for various catalytic reactions. All the carbon materials used as catalyst support constitute a large family of compounds that can vary both at textural and at structural levels. Today, the recent developments of well-controlled synthesis methodologies, advanced characterization techniques, and modeling tools allow one to correlate the relationships between metal/support/reactant at the molecular level. Based on these considerations, in this Review article, we wish to provide some answers to the question “How and why anchoring metal nanoparticles, clusters, or single atoms on carbon materials for catalysis?”. To do this, we will rely on both experimental and theoretical studies. We will first analyze what sites are available on the surface of a carbon support for the anchoring of the active phase. Then, we will describe some important effects in catalysis inherent to the presence of a carbon-type support (metal–support interaction, confinement, spillover, and surface functional group effects). These effects will be commented on by putting into perspective catalytic results obtained in numerous reactions of thermal catalysis, electrocatalysis, or photocatalysis.

Integration of electrochemical capacitors with silicon-based electronics is a major challenge, limiting energy storage on a chip. We describe a wafer-scale process for manufacturing strongly adhering carbide-derived carbon films and interdigitated micro-supercapacitors with embedded titanium carbide current collectors, fully compatible with current microfabrication and silicon-based device technology. Capacitance of those films reaches 410 farads per cubic centimeter/200 millifarads per square centimeter in aqueous electrolyte and 170 farads per cubic centimeter/85 millifarads per square centimeter in organic electrolyte. We also demonstrate preparation of self-supported, mechanically stable, micrometer-thick porous carbon films with a Young's modulus of 14.5 gigapascals, with the possibility of further transfer onto flexible substrates. These materials are interesting for applications in structural energy storage, tribology, and gas separation.

International audience

Glycogen and trehalose are the two glucose stores of yeast cells. The large variations in the cell content of these two compounds in response to different environmental changes indicate that their metabolism is controlled by complex regulatory systems. In this review we present information on the regulation of the activity of the enzymes implicated in the pathways of synthesis and degradation of glycogen and trehalose as well as on the transcriptional control of the genes encoding them. cAMP and the protein kinases Snf1 and Pho85 appear as major actors in this regulation. From a metabolic point of view, glucose-6-phosphate seems the major effector in the net synthesis of glycogen and trehalose. We discuss also the implication of the recently elucidated TOR-dependent nutrient signalling pathway in the control of the yeast glucose stores and its integration in growth and cell division. The unexpected roles of glycogen and trehalose found in the control of glycolytic flux, stress responses and energy stores for the budding process, demonstrate that their presence confers survival and reproductive advantages to the cell. The findings discussed provide for the first time a teleonomic value for the presence of two different glucose stores in the yeast cell.

Recent biotechnology advances allow for multiple types of omics data, such as transcriptomic, proteomic or metabolomic data sets to be integrated. The problem of feature selection has been addressed several times in the context of classification, but needs to be handled in a specific manner when integrating data. In this study, we focus on the integration of two-block data that are measured on the same samples. Our goal is to combine integration and simultaneous variable selection of the two data sets in a one-step procedure using a Partial Least Squares regression (PLS) variant to facilitate the biologists' interpretation. A novel computational methodology called ;;sparse PLS" is introduced for a predictive analysis to deal with these newly arisen problems. The sparsity of our approach is achieved with a Lasso penalization of the PLS loading vectors when computing the Singular Value Decomposition. Sparse PLS is shown to be effective and biologically meaningful. Comparisons with classical PLS are performed on a simulated data set and on real data sets. On one data set, a thorough biological interpretation of the obtained results is provided. We show that sparse PLS provides a valuable variable selection tool for highly dimensional data sets.

Monodispersed nanoparticles of cobalt have been prepared by an original method using the decomposition under hydrogen of an organometallic precursor in the presence of a stabilizing polymer. Two colloids (Coll-I and Coll-II) have been obtained by changing the organometallic concentration in the polymer. Observation by high-resolution transmission electronic microscopy (HRTEM) showed Co particles well isolated and regularly dispersed in the polymer with a very narrow size distribution centered around 1.5 nm (Coll-I) and 2 nm (Coll-II) diameter. These particles are superparamagnetic above the blocking temperature 9 K (Coll-I) and 13.5 K (Coll-II). The particle size deduced from the analyses of the magnetic susceptibilities and magnetization curves are consistent with those measured by HRTEM. Magnetization at 5 K seems to saturate in fields up to 5 T leading to an enhanced mean magnetic moment per atom for both samples, where $〈{\ensuremath{\mu}}_{\mathrm{Co}}〉=1.94\ifmmode\pm\else\textpm\fi{}0.05$ ${\ensuremath{\mu}}_{B}$ for the smallest particles. High-field magnetization measurements, up to 35 T, increases nearly linearly with the applied field. This is equivalent to an increase of the mean magnetic moment with $〈{\ensuremath{\mu}}_{\mathrm{Co}}〉=2.1\ifmmode\pm\else\textpm\fi{}0.1$ ${\ensuremath{\mu}}_{B}$ at 35 T for the smallest particles. The effective magnetic anisotropies are found to be larger than that of the bulk materials and decrease with increasing particle size. This set of data allows us to conclude that the enhanced magnetization, its increase with applied magnetic field, and the enhanced effective magnetic anisotropy are associated with the large influence of the surface atoms and are more significant with decreasing size.

Oxidative stress and resulting lipid peroxidation is involved in various and numerous pathological states including inflammation, atherosclerosis, neurodegenerative diseases and cancer. This review is focused on recent advances concerning the formation, metabolism and reactivity towards macromolecules of lipid peroxidation breakdown products, some of which being considered as 'second messengers' of oxidative stress. This review relates also new advances regarding apoptosis induction, survival/proliferation processes and autophagy regulated by 4-hydroxynonenal, a major product of omega-6 fatty acid peroxidation, in relationship with detoxication mechanisms. The use of these lipid peroxidation products as oxidative stress/lipid peroxidation biomarkers is also addressed.

Plastics are everywhere in our modern way of living, and their production keeps increasing every year, causing major environmental concerns. Nowadays, the end-of-life management involves accumulation in landfills, incineration, and recycling to a lower extent. This ecological threat to the environment is inspiring alternative bio-based solutions for plastic waste treatment and recycling toward a circular economy. Over the past decade, considerable efforts have been made to degrade commodity plastics using biocatalytic approaches. Here, we provide a comprehensive review on the recent advances in enzyme-based biocatalysis and in the design of related biocatalytic processes to recycle or upcycle commodity plastics, including polyesters, polyamides, polyurethanes, and polyolefins. We also discuss scope and limitations, challenges, and opportunities of this field of research. An important message from this review is that polymer-assimilating enzymes are very likely part of the solution to reaching a circular plastic economy.

In this paper, we address the issue of predicting the next location of an individual based on the observations of his mobility behavior over some period of time and the recent locations that he has visited. This work has several potential applications such as the evaluation of geo-privacy mechanisms, the development of location-based services anticipating the next movement of a user and the design of location-aware proactive resource migration. In a nutshell, we extend a mobility model called Mobility Markov Chain (MMC) in order to incorporate the n previous visited locations and we develop a novel algorithm for next location prediction based on this mobility model that we coined as n-MMC. The evaluation of the efficiency of our algorithm on three different datasets demonstrates an accuracy for the prediction of the next location in the range of 70% to 95% as soon as n = 2.

With continuing depletion of fresh water resources, focus has shifted more toward water recovery, reuse, and recycling, which require an extension of conventional wastewater treatment technologies.Downstream external factors like stricter compliance requirements for wastewater discharge, rising treatment costs, and spatial constraints necessitate renewed investigation of alternative technologies.Coupled with biological treatment processes, membrane technology has gained considerable attention due to its wide range of applicability and the performance characteristics of membrane systems that have been established by various investigations and innovations during the last decade.This article summarizes research efforts and presents a review of the how and why of their development and applications.The focus is on appraising and comparing technologies on the basis of their relative merits and demerits.Additional facts and figures, especially regarding process parameters and effluent quality, are used to evaluate primary findings on these technologies.Key factors such as loading rates, retention time, cross-flow velocities, membrane types, membrane fouling, and backwashing, etc. are some of the aspects covered.Membrane applications in various aerobic and anaerobic schemes are discussed at length.However, the emphasis is on the use of membranes as a solid/liquid separator, a key in achieving desired effluent quality.Further, technology development directions and possibilities are also explored.The review concludes with an economic assessment of the technologies because one of the key technology selection criteria is financial viability.

Constructal theory is the view that the generation of flow configuration is a physics phenomenon that can be based on a physics principle (the constructal law): “For a finite-size flow system to persist in time (to survive) its configuration must evolve in such a way that it provides an easier access to the currents that flow through it” [A. Bejan, Advanced Engineering Thermodynamics, 2nd ed. (Wiley, New York, 1997); Int. J. Heat Mass Transfer, 40, 799 (1997)]. This principle predicts natural configuration across the board: river basins, turbulence, animal design (allometry, vascularization, locomotion), cracks in solids, dendritic solidification, Earth climate, droplet impact configuration, etc. The same principle yields new designs for electronics, fuel cells, and tree networks for transport of people, goods, and information. This review describes a paradigm that is universally applicable in natural sciences, engineering and social sciences.