Portail HAL du Collège de France
Not a member yet
22122 research outputs found
Sort by
Evolution of a trait distributed over a large fragmented population: Propagation of chaos meets adaptive dynamics
No full textWe consider a metapopulation made up of demes, each containing individuals bearing a heritable quantitative trait. Demes are connected by migration and undergo independent Moran processes with mutation and selection based on trait values. Mutation and migration rates are tuned so that each deme receives a migrant or a mutant in the same slow timescale and is thus essentially monomorphic at all times for the trait (adaptive dynamics). In the timescale of mutation/migration, the metapopulation can then be seen as a giant spatial Moran model with size that we characterize. As and physical space becomes continuous, the empirical distribution of the trait (over the physical and trait spaces) evolves deterministically according to an integro-differential evolution equation. In this limit, the trait of every migrant is drawn from this global distribution, so that conditional on its initial state, traits from finitely many demes evolve independently (propagation of chaos). Under mean-field dispersal, the value of the trait at time and at any given location has a law denoted and a jump kernel with two terms: a mutation-fixation term and a migration-fixation term involving (McKean-Vlasov equation).In the limit where mutations have small effects and migration is further slowed down accordingly, we obtain the convergence of , in the new migration timescale, to the solution of a stochastic differential equation which can be referred to as a new canonical equation of adaptive dynamics. This equation includes an advection term representing selection, a diffusive term due to genetic drift, and a jump term, representing the effect of migration, to a state distributed according to its own law
Training of Physical Neural Networks
No full text29 pages, 4 figuresInternational audiencePhysical neural networks (PNNs) are a class of neural-like networks that leverage the properties of physical systems to perform computation. While PNNs are so far a niche research area with small-scale laboratory demonstrations, they are arguably one of the most underappreciated important opportunities in modern AI. Could we train AI models 1000x larger than current ones? Could we do this and also have them perform inference locally and privately on edge devices, such as smartphones or sensors? Research over the past few years has shown that the answer to all these questions is likely "yes, with enough research": PNNs could one day radically change what is possible and practical for AI systems. To do this will however require rethinking both how AI models work, and how they are trained - primarily by considering the problems through the constraints of the underlying hardware physics. To train PNNs at large scale, many methods including backpropagation-based and backpropagation-free approaches are now being explored. These methods have various trade-offs, and so far no method has been shown to scale to the same scale and performance as the backpropagation algorithm widely used in deep learning today. However, this is rapidly changing, and a diverse ecosystem of training techniques provides clues for how PNNs may one day be utilized to create both more efficient realizations of current-scale AI models, and to enable unprecedented-scale models
Splice-altering variant of PJVK gene in a Mauritanian family with non-syndromic hearing impairment
No full textInternational audiencePJVK gene was recently shown to create hypervulnerability to sound in humans and was the first human gene implicated in non-syndromic hearing impairment due to neural defect. Targeted next-generation sequencing of over 150 known deafness genes was performed in the proband. Sanger sequencing was used to validate the PJVK variant and confirm familial segregation of the disease. A minigene-based assay has been performed to assess the impact of the variant on splicing. We identified a novel c.550-6A > G acceptor splice-site variant in the PJVK gene in the homozygous state in a Mauritanian child with severe to profound congenital deafness. The substitution was located in intron 4. The effect of the variation was demonstrated by a minigene assay which showed that the variation, an insertion of an additional 5 bp, created a new splice site resulting in the appearance of a premature stop codon (p.Phe184Tyrfs*26) and likely a truncated protein. This result constitutes a new splice-site variant report in the PJVK gene leading to DFNB59 type associated with autosomal recessive non-syndromic hearing impairment (ARNSHI)
Temporalities in Mesoamerican Ritual Practices: An Introduction
No full textInternational audienceTemporalities in Mesoamerican Ritual Practices examines the time-based dimensions of ritual activities in past and present Mesoamerican societies, including the prehispanic, colonial, and modern periods. The authors explore ritual around three principal categories of action—creating, transforming, and destroying—as significant cultural manifestations of the temporal dimension of transition processes. Based on specific case studies, new analysis of fieldwork data, and long-term collaboration between authors, chapters engage empirically and theoretically with the multiple temporalities of ritual in relation to both the unfolding of ritual performance and its external and symbolic anchors. Taking rituals as a series of specific formalized actions that produce transitory changes within an initial context, the authors examine activities that generate change linked to artifact production, life cycles, healing, conflict resolution, crisis management, the enthronement of rulers and transfers of responsibilities, and practices relating to the occupation, abandonment, reuse, or conversion of socialized spaces. Adopting a multidisciplinary approach in archaeology, ethnohistory, anthropology, and linguistic anthropology, Temporalities in Mesoamerican Ritual Practices offers new insights into ritual time approached through multi-semiotic, material, sensorial, and pragmatic perspectives that encourage further interdisciplinary dialogue. Contributors: Chloé Andrieu, Charlotte Arnauld, Cédric Becquey, Johan Begel, Alain Breton, Marie Chosson, Flavia G. Cuturi, Daniel Dehouve, Bérénice Gaillemin, Alex Geurds, Rosemary A. Joyce, Leonardo López Luján, Alessandro Lupo, Tsubasa Okoshi, Divina Perla Barrera, Perig Pitrou, Sergio Romero, Julien Sion, Carlos Efraín Tox, Fidel Tuyuc Ni
Table ronde "Paroles de Sciences". Quelles recherches pour affronter les crises et penser l'avenir ?
No full textInternational audienc
BISANZIO E L’OCCIDENTE NELL’XI SECOLO: NUOVE SFIDE, NUOVE OPPORTUNITÀ
No full textInternational audienc
Le discours d’ouverture de Nicolas Iorga au congrès de 1924. Byzance entre réminiscences et rémanences
No full textInternational audienceThe organization of the First International Congress of Byzantine Studies in Bucharest owed much to the personal initiative of the Romanian historian Nicolae Iorga. The congress opened on Monday, 14 April 1924, with Iorga, as president of the organizing committee, welcoming the delegates from across Europe. This article analyzes his inaugural address, highlighting Iorga’s understanding of the Byzantine legacy, the traces of that legacy preserved in Romanian culture, and the place of Byzantine and post‑Byzantine history within universal history
