HAL-Artois
Not a member yet
27242 research outputs found
Sort by
Enseigner la géographie avec l’Intelligence Artificielle, approches expérientielles
No full textInternational audienceL’usage de l’IA dans l’enseignement de la géographie est encore balbutiante. Dans le cadre du groupe Pensée Spatiale de l’IREMS de Paris des usages ont été testés en mobilisant une approche expérientielle, en formation d’enseignants et dans la classe.Cette communication propose de rendre compte des possibles explorés autour de trois expérimentations. La première porte sur la formation d’enseignants en formation initiale dans le cadre d’une UE de pré-professionnalisation où l’IA a été utilisée pour préparer une séquence. La seconde porte sur l’usage de l’IA pour faire raisonner les élèves en créant des bandes dessinées avec l’IA en collège. La troisième porte sur les enjeux des trajectoires démographiques différenciées dans le monde en lycée où les élèves ont préparé un débat à partir d’interaction avec un Chabot
Improving B2B customer churn through action rule mining
No full textInternational audienceBusiness-to-business (B2B) firms must maintain robust customer bases to ensure recurring revenue. To do so effectively, they should engage in churn prediction. Proactively identifying potential churners and taking proactive retention measures help companies safeguard their revenue streams and build strong, long-lasting relationships with customers, which enhances their sustainability and competitive performance in dynamic, competitive markets. Yet, extant B2B customer churn models often fail to offer truly practical or actionable decision support, such that marketers must rely on their intuition and exert additional effort to define appropriate preventive retention measures. To address this research gap between research models and actionable insights, the current study proposes B2B-ARM, a B2B actionable rule model (ARM), that offers clear action paths for proactive retention management. A real-life case study of a European B2B software company with 6275 contracts provides benchmark evidence that B2B-ARM can detect churn equally well as popular existing prediction models (i.e., decision tree, logistic regression, and naïve Bayes). Furthermore, B2B-ARM provides actionable recommendations, as well as direct remedies to prevent churn, such that marketers save both time and resources. Overall, B2B-ARM is a reliable, efficient, and practical tool for mitigating B2B churn and improving customer retention
Bridging Institutional Theory and Social and Environmental Efforts in Management: A Review and Research Agenda
No full textInternational audienceThis review examines the integration of institutional theory with social and environmental efforts in management (i.e., regarding sustainability, corporate social responsibility, and environmental, social, and governance objectives). By analyzing 720 studies published between 1997 and 2023, we develop a multi-level model that maps the antecedents of different actors (e.g., industries, organizations, individuals) to respond to or reshape institutional structures, the mechanisms they use, the moderators, and outcomes of their practices. Our findings emphasize the dynamic interplay between structure and agency across systemic, organizational, and individual levels, offering a comprehensive framework for future research. We highlight three key observations: first, while substantial research explores how institutions shape actors, more attention is needed to understand the reciprocal influences as actors are shaped by and reshape institutions over time. Second, individual-level dynamics remain significantly underexplored, with limited focus on resistance, demotivation, and failure—essential elements of the complexity of institutional processes. Finally, we identify a critical need to examine the unintended consequences of social and environmental efforts, revealing how these endeavors may undermine their goals, create new challenges, or generate unexpected solutions
Transforming spontaneous premature neonatal EEG to spontaneous fetal MEG using a novel machine learning approach
No full textInternational audienceObjectivesThe spontaneous neural activity of premature neonates has been characterized with electroencephalography (EEG). However, evaluation of normal and pathological fetal brain development is still largely unknown. Fetal magnetoencephalography (fMEG) is currently the only available technique to record fetal neural activity. Benefiting from progress in machine learning and artificial intelligence, we aimed to transfer premature EEG to fMEG, to characterize the manifestation of spontaneous activity using the knowledge obtained from premature EEG.MethodsIn this study, 30 high-resolution EEG recordings from premature newborns and 44 fMEG recordings were used to develop a transfer function to predict the spontaneous neural activity of the fetus. After preprocessing, bursts of spontaneous activity were detected using the non-linear energy operator. Next, we proposed a CycleGAN-based model to transform the premature EEG to fMEG and evaluated its performance with both time and frequency measurements.ResultsIn the time domain, the values were similar for the mean square error (< 5 %) and correlation (0.91 ± 0.05 and 0.89 ± 0.08) for both transformations between the original data and that generated by CycleGAN. However, considering the frequency content, the CycleGAN-based model modulated the frequency content of EEG to MEG transformed signals relative to the original signals by increasing the power, on average, in all frequency bands, except for the slow delta frequency band.ConclusionOur developed model showed promising potential to generate a priori signatures of fMEG manifestations related to spontaneous neural activity. Collectively, this study represents the first steps toward identifying neurobiomarkers of fetal brain development
Upcycling Waste Polycarbonate into <i>N,N</i>'-Diphenylethylurea: A Hands-On Experiment for Undergraduate Chemistry Laboratories
No full textInternational audienceThe urgent challenge of climate change drives the need for sustainable and environmentally friendly practices, such as recycling, upcycling, and sustainable chemistry. Despite their importance, these practices are rarely integrated into undergraduate chemistry laboratories. Here, we describe a hands-on approach suitable for a second-semester organic chemistry lab that incorporates waste management principles, guiding students to recognize the value of repurposing chemical waste. The precursors for this experiment are discarded CDs and DVDs, which are readily available in most households. This not only helps reduce waste but also demonstrates how everyday materials can be transformed into high-value chemicals (HVCs). In this lab, students extract a carbonyl group from old CDs and DVDs, using a process called “carbonyl harvesting,” to show how waste can be turned into valuable materials. The recycled bisphenol A (BPA) is monitored using thin layer chromatography (TLC) and collected separately, while the N,N’-diphenylethylurea is purified by simple filtration and characterized using 1H, 13C NMR, IR, XRD, and melting point analysis. This approach not only reinforces fundamental organic lab techniques but also provides hands-on training in sustainability lab practices, showing students how to integrate waste management, resource recovery, and sustainability into their experimental work. Its versatility allows it to be integrated into chemical engineering, general chemistry, and biotechnology curricula. It serves as a valuable learning tool by demonstrating sustainable process design, functional group preservation, and selective chemical transformations
Development of a test to compare working memory and problem-solving abilities in young and aged healthy cats -the felicog project
No full textInternational audienc
Harnessing the power of anti-amyloidogenic polyglutamine binding peptide 1: A computational and biophysical approach to antimicrobial prediction
No full textInternational audienceIn this work, we demonstrate for the first time the antimicrobial activity of poly-Q binding peptide 1 (QBP1), an anti-amyloidogenic molecule previously identified by phage display for its ability to bind and inhibit the aggregation of polyglutamine proteins like Huntingtin, but also α-synuclein and prion models. Intriguingly, sequence analysis by the ADAPTABLE web server highlighted QBP1's potential antifungal and antibacterial activity, which we have now confirmed experimentally. A theoretical basis for the predicted mechanism of action was provided by molecular dynamics simulations revealing the role of QBP1 aggregates promoting membrane disruption and specific interactions between QBP1 and bacterial and fungal phospholipids, further substantiated by solid-state NMR studies. Using primary human cells, we demonstrated that QBP1 does not display toxicity at high concentrations. Finally, our data demonstrate QBP1's efficacy against some strains of Bacillus cereus, B. mojavensis, Staphylococcus aureus, S. epidermidis, Micrococcus luteus, Enterococcus faecalis, Escherichia coli, and Candida. This unexpected dual function of QBP1 opens new avenues for therapeutic development, potentially restoring the putative antimicrobial protection exerted by many amyloidogenic proteins
Fracture-induced variations in mechanical properties of silty mudstone under triaxial stress
No full textInternational audienc
A glycosylated lipooctapeptide promotes uptake and growth of Mycobacterium abscessus in the host
No full textInternational audiencePathogenic mycobacteria produce a wide array of lipids which participate in host cell interactions and virulence. While some of these are conserved across all mycobacteria, others, like glycopeptidolipids (GPL), are restricted to a few species. Mycobacterium abscessus , an emerging rapid-growing pathogen, transitions from a smooth to a virulent rough variant upon the loss of surface GPL. Here, we discovered that M. abscessus and phylogenetically-close species harbor a second GPL-related locus, comprising two adjacent non-ribosomal peptide synthetase genes, MAB_4690c and MAB_4691c . A MAB_4690c deletion mutant (Δ MAB_4690c ) failed to produce a yet undescribed lipid, designated GL8P for glycosylated lipooctapeptide, sharing an acylated octapeptide core adorned by mono or di- O -rhamnosyl substituents. Δ MAB_4690c exhibited impaired uptake and survival in THP-1 cells and was attenuated in mice. Importantly, GL8P elicited a strong humoral response in patients infected with M. abscessus . These results highlight the role of GL8P in the pathophysiology of infection by rough M . abscessus and suggest its potential as a selective marker for M . abscessus infections