École Polytechnique Fédérale de Lausanne
Infoscience - École polytechnique fédérale de LausanneNot a member yet
191401 research outputs found
Sort by
SARS-CoV-2 Infection and the Risk of New Chronic Conditions: Insights from a Longitudinal Population-Based Study
No full textBackground: The post-acute impact of SARS-CoV-2 infections on chronic conditions remains poorly understood, particularly in general populations. Objectives: Our primary aim was to assess the association between SARS-CoV-2 infections and new diagnoses of chronic conditions. Our two secondary aims were to explore geographic variations in this association and to assess the association between SARS-CoV-2 infections and the exacerbation of pre-existing conditions. Methods: This longitudinal study used data from 8086 participants of the Specchio-COVID-19 cohort in the canton of Geneva, Switzerland (2021–2023). Mixed-effects logistic regressions and geographically weighted regressions adjusted for sociodemographic, socioeconomic, and healthcare access covariates were used to analyze self-reported SARS-CoV-2 infections, new diagnoses of chronic conditions, and the exacerbation of pre-existing ones. Results: Participants reporting a SARS-CoV-2 infection were more likely to be diagnosed with a new chronic condition compared to those who did not report an infection (adjusted odds ratio [aOR] = 2.15, 95% CI 1.43–3.23, adjusted p-value = 0.002). Notable geographic variations were identified in the association between SARS-CoV-2 infections and new diagnoses. While a positive association was initially observed between SARS-CoV-2 infections and exacerbation of pre-existing chronic conditions, this association did not remain significant after adjusting p-values for multiple comparisons. Conclusions: These findings contribute to understanding COVID-19’s post-acute impact on chronic conditions, highlighting the need for targeted health management approaches and calling for tailored public health strategies to address the pandemic’s long-term effects.LG
How Do Multilingual Language Models Remember Facts?
No full textLarge Language Models (LLMs) store and retrieve vast amounts of factual knowledge acquired during pre-training. Prior research has localized and identified mechanisms behind knowledge recall; however, it has only focused on English monolingual models. The question of how these mechanisms generalize to non-English languages and multilingual LLMs remains unexplored. In this paper, we address this gap by conducting a comprehensive analysis of three multilingual LLMs. First, we show that previously identified recall mechanisms in English largely apply to multilingual contexts, with nuances based on language and architecture. Next, through patching intermediate representations, we localize the role of language during recall, finding that subject enrichment is language-independent, while object extraction is language-dependent. Additionally, we discover that the last token representation acts as a Function Vector (FV), encoding both the language of the query and the content to be extracted from the subject. Furthermore, in decoder-only LLMs, FVs compose these two pieces of information in two separate stages. These insights reveal unique mechanisms in multilingual LLMs for recalling information, highlighting the need for new methodologies-such as knowledge evaluation, fact editing, and knowledge acquisition-that are specifically tailored for multilingual LLMs.NLPLSI
Cultural Heritage and Urban Landscapes: an Anthropological Approach
No full textThis article examines the anthropological approach to study heritage in urban landscapes, focusing on how cultural assets are constructed, preserved, and contested. The discussion situates heritage within evolving institutional frameworks, questioning the criteria for heritage designation and its implications for communities. A key focus is the "anthropologisation" of culture – a trend supported by UNESCO conventions that emphasise intangible heritage and community participation. Policy frameworks such as the Historic Urban Landscape (HUL) approach and ICOMOS charters exemplify a shift from a universalist perspective on heritage to more culturally relativist viewpoints. The case study of Beijing's Gulou neighbourhood illustrates the complexities of heritage preservation in rapidly evolving urban environments, underscoring community resistance and alternative memory practices. This article also discusses the emerging interdisciplinary field of heritage sciences, where social sciences and technology converge to redefine conservation practices. By analysing the intersections of heritage, memory, and policy, it enriches discussions on cultural identity and the evolving role of heritage in contemporary urban landscapes.SCI-ENAC-FG
The status of refrigeration techniques for vaccine storage and transportation in low-income settings
No full textVaccines need to be continuously stored between 2°C to 8°C, from their production to administration to beneficiaries. Every year, more than 25% of vaccines are wasted. One of the main reasons for this wastage is the lack of cold chain continuity in low-income settings, where electricity is scarce. Recently, several advances have been made in cooling technologies to store and transport vaccines. The current paper presents a review of refrigeration technologies based on scientific publications, industry white papers and other grey literature. For each refrigeration method, we describe its working principle, the best performing devices available as well as the remaining research challenges in order to obtain a very high degree of performance enhancement. Finally, we comment on their applicability for vaccine transport and storage.ESSTECH-G
Graph Neural Networks With Adaptive Structures
No full textGraph neural networks (GNNs) have made significant progress in various machine learning tasks. Despite their success, many existing GNN models are shown to be vulnerable to adversarial attacks, creating a stringent need to build robust GNN architectures. In this work, we introduce a novel interpretable message passing scheme over adaptive graph structures (ASMP) to enhance the resilience of GNNs against graph structural attacks. The ASMP layers are constructed through optimization steps that concurrently optimize node features and graph structures, allowing the message passing process to be conducted across dynamically adjusted graphs at different layers. This adaptability enables ASMP to handle noisy or perturbed graph structures more effectively, enhancing robustness. We also establish the theoretical convergence properties for the ASMP scheme. By integrating ASMP with neural networks, we introduce a new class of GNNs with adaptive structures (ASGNNs). Extensive experiments on semi-supervised node classification tasks demonstrate that ASGNN outperforms the state-of-the-art GNN architectures regarding classification accuracy when subjected to various adversarial attack scenarios.IMO
The Odyssey of Commonsense Causality: From Foundational Benchmarks to Cutting-Edge Reasoning
No full textUnderstanding commonsense causality is a unique mark of intelligence for humans. It helps people understand the principles of the real world better and benefits the decisionmaking process related to causation. For instance, commonsense causality is crucial in judging whether a defendant's action causes the plaintiff's loss in determining legal liability. Despite its significance, a systematic exploration of this topic is notably lacking. Our comprehensive survey bridges this gap by focusing on taxonomies, benchmarks, acquisition methods, qualitative reasoning, and quantitative measurements in commonsense causality, synthesizing insights from over 200 representative articles. Our work aims to provide a systematic overview, update scholars on recent advancements, provide a pragmatic guide for beginners, and highlight promising future research directions in this vital field. A summary of the related literature is available at https://github. com/cui-shaobo/causality-papers .LI
First Evidence for Direct CP Violation in Beauty to Charmonium Decays
No full textThe CP asymmetry and branching fraction of the Cabibbo-Kobayashi-Maskawa-suppressed decay B+→J/ψπ+ are precisely measured relative to the favored decay B+→J/ψK+ using a sample of proton-proton collision data corresponding to an integrated luminosity of 5.4 fb-1 recorded at a center-of-mass energy of 13 TeV during 2016-2018. The results of the CP asymmetry difference and branching fraction ratio are ΔACPACP(B+→J/ψπ+)-ACP(B+→J/ψK+)=(1.29±0.49±0.08)×10-2, Rπ/K[B(B+→J/ψπ+)/B(B+→J/ψK+)]=(3.852±0.022±0.018)×10-2, where the first uncertainties are statistical and the second are systematic. A combination with previous LHCb results based on data collected at 7 and 8 TeV in 2011 and 2012 yields ΔACP=(1.42±0.43±0.08)×10-2 and Rπ/K=(3.846±0.018±0.018)×10-2. The combined ΔACP value deviates from zero by 3.2 standard deviations, providing the first evidence for direct CP violation in the amplitudes of beauty decays to charmonium final states.LPHE-OSPH-SBLPHE-RMLPHE-LSIN-
Cortical and behavioral tracking of rhythm in music: Effects of pitch predictability, enjoyment, and expertise
No full textThe cortical tracking of stimulus features is a crucial neural requisite of how we process continuous music. We here tested whether cortical tracking of the beat, typically related to rhythm processing, is modulated by pitch predictability and other top-down factors. Participants listened to tonal (high pitch predictability) and atonal (low pitch predictability) music while undergoing electroencephalography. We analyzed their cortical tracking of the acoustic envelope. Cortical envelope tracking was stronger while listening to atonal music, potentially reflecting listeners' violated pitch expectations and increased attention allocation. Envelope tracking was also stronger with more expertise and enjoyment. Furthermore, we showed cortical tracking of pitch surprisal (using IDyOM), which suggests that listeners' expectations match those computed by the IDyOM model, with higher surprisal for atonal music. Behaviorally, we measured participants' ability to finger-tap to the beat of tonal and atonal sequences in two experiments. Finger-tapping performance was better in the tonal condition, indicating a positive effect of pitch predictability on behavioral rhythm processing. Cortical envelope tracking predicted tapping performance for tonal music, as did pitch-surprisal tracking for atonal music, indicating that high and low predictability might impose different processing regimes. Taken together, our results show various ways that top-down factors impact musical rhythm processing.DCM
The polyphenol metabolite urolithin A suppresses myostatin expression and augments glucose uptake in human skeletal muscle cells
No full textPurpose Polyphenolic plant extracts have demonstrated anti-inflammatory and anti-catabolic effects in vitro, however their meaningful translation into humans remains elusive. Urolithin A (UA), a gut-derived metabolite of ellagitannins, has shown promise for improving muscle function and metabolic health in rodent models. This study aimed to explore the impact of UA on insulin and anabolic sensitivity in human skeletal muscle cells. Methods Primary human myogenic cultures were derived from skeletal muscle biopsies of eight healthy adults. After differentiation, myotubes were treated with 0.002, 1 and 50 mu M UA or vehicle for 24 h. Cell viability was assessed using a resazurin assay. Basal and insulin-stimulated glucose uptake was measured using tritiated deoxy-D-glucose, whilst amino acid-stimulated protein synthesis was estimated using the surface sensing of translation (SuNSET) technique. Expression of myostatin and glucose transporters was quantified via real-time PCR. Results UA treatment at M UA reduced myostatin (MSTN) expression by 14% (P < 0.01) but did not alter amino acid-stimulated global cell protein synthesis. Conclusion This study provides evidence of UA's metabolic benefits in primary human myotubes, notably improving basal- and insulin-stimulated glucose uptake and supressing MSTN expression. These findings suggest UA could be an effective nutraceutical for mitigating insulin resistance and warrants further investigation.Non-EPF