A5044018048- Neural Networks and Applications
- Neural dynamics and brain function
- COVID-19 epidemiological studies
- Phonetics and Phonology Research
- Music and Audio Processing
- Cell Image Analysis Techniques
- Cognitive Science and Education Research
- Music Technology and Sound Studies
- Genomic variations and chromosomal abnormalities
- Language and cultural evolution
- Linguistic Variation and Morphology
- Second Language Acquisition and Learning
- Fractal and DNA sequence analysis
- Genetics and Neurodevelopmental Disorders
- Neuroscience and Music Perception
- Yersinia bacterium, plague, ectoparasites research
- Autism Spectrum Disorder Research
- Hydrology and Drought Analysis
- COVID-19 Clinical Research Studies
- Multisensory perception and integration
- Zoonotic diseases and public health
- Animal Vocal Communication and Behavior
- Advanced Neural Network Applications
- Face Recognition and Perception
- Sepsis Diagnosis and Treatment
Centre National de la Recherche Scientifique
2008-2024
Centre d'Analyse et de Mathématique Sociales
2008-2024
Laboratoire de Recherche en Informatique
2022
Université Paris-Saclay
2022
École Normale Supérieure
2022
École Normale Supérieure d'Abidjan
2022
Google (United States)
2022
Institut national de recherche en informatique et en automatique
2022
École des hautes études en sciences sociales
2008-2021
Université Paris Sciences et Lettres
2018
MRI has been extensively used to identify anatomical and functional differences in Autism Spectrum Disorder (ASD). Yet, many of these findings have proven difficult replicate because studies rely on small cohorts are built complex, undisclosed, analytic choices. We conducted an international challenge predict ASD diagnosis from data, where we provided preprocessed data > 2,000 individuals. Evaluation the predictions was rigorously blinded. 146 challengers submitted prediction algorithms,...
Abstract As a large-scale instance of dramatic collective behaviour, the 2005 French riots started in poor suburb Paris, then spread all France, lasting about three weeks. Remarkably, although there were no displacements rioters, riot activity did travel. Access to daily national police data has allowed us explore dynamics propagation. Here we show that an epidemic-like model, with just few parameters and single sociological variable characterizing neighbourhood deprivation, accounts...
There has been a long history of works showing that neural networks have hard time extrapolating beyond the training set. A recent study by Balestriero et al. (2021) challenges this view: defining interpolation as state belonging to convex hull set, they show test either in input or space, cannot lie for most part hull, due high dimensionality data, invoking well known curse dimensionality. Neural are then assumed necessarily work extrapolative mode. We here activities last hidden layer...
Languages have long been described according to their perceived rhythmic attributes. The associated typologies are of interest in psycholinguistics as they partly predict newborns' abilities discriminate between languages and provide insights into how adult listeners process non-native languages. Despite the relative success rhythm metrics supporting existence linguistic classes, quantitative studies yet capture full complexity temporal regularities with speech rhythm. We argue that deep...
Background: Reliable information is an essential component for responding to the COVID-19 epidemic, especially regarding availability of critical care beds (CCBs). We propose three contributions: a) ICUBAM (ICU Bed Availability Monitor), a tool which both collects and visualizes on CCB entered directly by intensivists. b) An analysis ICU admissions outcomes using collected \ICUBAM during 6-week period in hard-hit Grand Est region France, c) Explanatory predictive models adapted prediction,...
The state of the art in many computer vision tasks is represented by Convolutional Neural Networks (CNNs). Although their hierarchical organization and local feature extraction are inspired structure primate visual systems, lack lateral connections such architectures critically distinguishes analysis from biological object processing. idea enriching CNNs with recurrent convolutional type has been put into practice recent years, form learned kernels no geometrical constraints. In present...
Classification is one of the major tasks that deep learning successfully tackling. Categorization also a fundamental cognitive ability. A well-known perceptual consequence categorization in humans and other animals, categorical perception, notably characterized by within-category compression between-category separation: two items, close input space, are perceived closer if they belong to same category than different categories. Elaborating on experimental theoretical results science, here we...
Abstract Since its inception, Shannon's information theory has attracted interest for the study of language and music. Recently, a wide range converging studies have shown how efficient communication pervades language, from phonetics to syntax. Efficient principles imply that more resources should be assigned highly informative items. For instance, average content was better predictor word length than frequency, revisiting famous Zipf's law. However, in spite success framework speech, very...
Sensory systems are permanently bombarded with complex stimuli. Cognitive processing of such stimuli may be facilitated by accentuation important elements. In the case music listening, alteration some surface features –such as volume and duration– facilitate cognitive otherwise high-level information, melody harmony. Hence, musical accents often aligned intrinsically salient elements in stimuli, highly unexpected notes. We developed a novel listening paradigm based on an artificial...
Abstract MRI has been extensively used to identify anatomical and functional differences in Autism Spectrum Disorder (ASD). Yet, many of these findings have proven difficult replicate because studies rely on small cohorts are built complex, undisclosed, analytic choices. We conducted an international challenge predict ASD diagnosis from data, where we provided preprocessed data > 2,000 individuals. Evaluation the predictions was rigorously blinded. 146 challengers submitted prediction...
Categorization is an important topic both for biological and artificial neural networks. Here, we take information theoretic approach to assess the efficiency of representations induced by category learning. We show that one can decompose relevant Bayesian cost into two components, coding part decoding part. Minimizing implies maximizing mutual between set categories activities. analytically this be written as sum terms interpreted (i) finding appropriate representation space, and, (ii)...
Motivated by traditional rhythm class typologies, studies of language-learners’ typically focus on the syllable or segment level. Studying word-level lets us explore effects lexical features (e.g., part speech, predictability) word durations in non-native speech. This study examined whether native and English can be distinguished variation realization features, non-native-like leads to a stronger foreign accent. Word were measured paragraphs read 12 American (AE), 20 Korean, Chinese...
This paper takes an information-theoretic approach to study the perceptual consequences of neural encoding categories (e.g. vowels). We focus on two well-known psychophysical phenomena: categorical perception, characterized by greater cross-category than within-category discrimination, and magnet effect, stating that space is wrapped around prototypical instances a given category, leading better ability discriminate stimuli near non-prototypical exemplars category ones. Introducing distance...
Phonetic categories are known to induce good discrimination of stimuli from different and poor within category [Liberman et al., J. Exp. Psych. 54, 358–368 (1957)]. In addition, passive exposure a phonetic continuum improves if presented according bimodal distribution occurrence frequency, whereas unimodal frequency results in decreased discriminability [Maye Dev. Sci. 11, 122–134 (2008)]. The present study examined whether versus distributions categorical pattern perception. Stimuli were...
Reliable information is an essential component of responding to a sudden and large disease outbreak such as COVID-19, particularly with respect critical care beds (CCBs) availability. This article presents: i) the development construction ICUBAM, tool that collects in real-time visualizes on CCB availability entered directly by intensivists; ii) analysis interpretation data collected over 6-week period during first wave epidemic hard-hit Grand Est region France; iii) region; iv) medium long...
Abstract Somax2 is an artificial intelligence (AI)-based multiagent system for human–machine “coimprovisation” that generates stylistically coherent streams while continuously listening and adapting to musicians or other agents. The model on which it based can be used with little configuration interact humans in full autonomy, but also allows fine real-time control of its generative processes interaction strategies, closer this case a “smart” digital instrument. An offspring the Omax system,...