HAL-Polytechnique
Not a member yet
51406 research outputs found
Sort by
Efficient Crawling for Scalable Web Data Acquisition
No full textInternational audienceJournalistic fact-checking, as well as social or economic research, require analyzing high-quality statistics datasets (SDs, in short). However, retrieving SD corpora at scale may be hard, inefficient, or impossible, depending on how they are published online. To improve open statistics data accessibility, we present a focused Web crawling algorithm that retrieves as many targets, i.e., resources of certain types, as possible, from a given website, in an efficient and scalable way, by crawling (much) less than the full website. We show that optimally solving this problem is intractable, and propose an approach based on reinforcement learning, namely using sleeping bandits. We propose SB-CLASSIFIER, a crawler that efficiently learns which hyperlinks lead to pages that link to many targets, based on the paths leading to the links in their enclosing webpages. Our experiments on websites with millions of webpages show that our crawler is highly efficient, delivering high fractions of a site's targets while crawling only a small part
Modular abelian surfaces of small conductor with nontrivial Tate--Shafarevich groups
No full textWe exhibit examples of geometrically simple abelian surfaces A/ℚ with conductor bounded by (10 000)2 whose Tate—Shafarevich groups contain a subgroup isomorphic to (ℤ/pℤ)2 for each p = 5, 7, 11, 13. To find these examples we generalise work of Cremona–Freitas to enumerate all congruences of a certain type between pairs of weight 2 newforms f ∈ S2new(Γ0(N)) and g ∈ S2new(Γ0(M)) contained in the LMFDB (i.e., with N, M < 10 000) and with coefficient fields of degree ≤ 4. Passing from the modular forms to the corresponding abelian varieties we use visibility to (unconditionally) prove the existence of non-trivial elements of the Tate–Shafarevich group. Finally we construct an example of an abelian surface with (ℤ/7ℤ)2 ⊂ Ш(A/ℚ) which is (conjecturally) not visible in any abelian threefold.</div
Pentanucleotide guanine-rich WGGGW repeats, including CANVAS AGGGA repeats, form a variety of noncanonical structures
No full textInternational audienceAbstract Short tandem repeats (STRs) are an important component of the human genome as they contribute to genetic diversity and can influence gene expression and disease susceptibility. STRs are important in the context of CANVAS (Cerebellar Ataxia, Neuropathy, Vestibular Areflexia Syndrome) genetic disease as expansions of AGGGA repeats within the RFC1 gene are associated with the development of this neurodegenerative disorder. Interestingly, the RFC1 expanded motifs are pentanucleotides that differ from the nonpathogenic AGAAA pentanucleotide motif present in reference genomes. The molecular mechanisms underlying the pathogenicity of the mutated pentanucleotide expansion in CANVAS are still unknown. Several groups have shown that DNA and RNA containing AGGGA repeats fold into G-quadruplexes (G4s) under physiological K⁺ conditions. In this study, we reveal a more complex than expected behavior, in which DNA WGGGW motifs (where W is A or T) may adopt different G4 and non-G4 structures depending on sequence, repeat number and ionic conditions. These findings are relevant as they may help explain the genomic instability and pathogenicity specifically associated with AGGGA repeats among the WGGGW motifs
Peace Talk and Conflict Traps
No full textNational audienceCostly pre-play messages can deter unnecessary wars—but the same messages can also entrench stalemates once violence begins. We develop an overlapping-generations model of a security dilemma with persistent group types (normal vs. bad), one-sided private signaling by the current old to the current young, and noisy private memory of the last encounter. We characterize a stationary equilibrium in which, for an intermediate band of signal costs, normal old agents mix on sending a costly reassurance only after an alarming private history; the signal is kept marginally persuasive by endogenous receiver cutoffs and strategic mimicking by bad types. Signaling strictly reduces the hazard of conflict onset; conditional on onset, duration is unchanged in the private model but increases once a small probability of publicity (leaks) creates a public record of failed reconciliation. With publicity, play generically absorbs in a peace trap or a conflict trap. We discuss welfare and policy: when to prefer back-channels versus public pledges
Efficient Crawling for Scalable Web Data Acquisition (Extended Version)
No full textJournalistic fact-checking, as well as social or economic research, require analyzing high-quality statistics datasets (SDs, in short). However, retrieving SD corpora at scale may be hard, inefficient, or impossible, depending on how they are published online. To improve open statistics data accessibility, we present a focused Web crawling algorithm that retrieves as many targets, i.e., resources of certain types, as possible, from a given website, in an efficient and scalable way, by crawling (much) less than the full website. We show that optimally solving this problem is intractable, and propose an approach based on reinforcement learning, namely using sleeping bandits. We propose SB-CLASSIFIER, a crawler that efficiently learns which hyperlinks lead to pages that link to many targets, based on the paths leading to the links in their enclosing webpages. Our experiments on websites with millions of webpages show that our crawler is highly efficient, delivering high fractions of a site's targets while crawling only a small part
Measurement of the (1S), (2S), and (3S) differential cross sections in pp collisions at = 13.6 TeV
No full textInternational audienceThe production cross sections of the (1S), (2S), and (3S) mesons are measured in proton-proton collisions at = 13.6 TeV, using a data sample collected in 2022 by the CMS experiment and corresponding to an integrated luminosity of 37.4 fb. The measurement is performed in the decay channels, differentially as a function of transverse momentum in the 20200 GeV range, in the 0.6 and 0.6 1.2 rapidity intervals
Improved well-posedness for the limit flow of differentiation of roots of polynomials
No full textIn this paper, we study the partial differential equation on the circle that was heuristically obtained by Steinerberg [32] on the real line and which represents the evolution of the density of the roots of polynomials under differentiation. After integrating the partial differential equation in question, we observe that it can be treated with the theory of viscosity solutions. This equation at hand is a non linear parabolic integro-differential equation which involves the elliptic operator called the half-Laplacian. Due to the singularity of the equation, we restrict our study to strictly positive initial condition. We obtain a comparison principle for solutions of the primitive equation which yields uniqueness, existence, continuity with respect to initial condition. We also present heuristics to justify that the system of particles indeed approximates the solution of the equation
D-LIM: a Neural Network for Interpretable Gene-Gene Interactions
No full textRecent advances in gene editing can produce large genotype-fitness maps for targeted genes, yet predicting the effects of mutations between genes remains challenging. Indeed, biochemical models require knowledge of underlying parameters and interactions, whereas machine learning methods typically lack interpretability, as they do not link model parameters to biological quantities. We introduce D-LIM, a neural network that infers low-dimensional fitness landscapes directly from mutation-fitness data. The distinctive feature of D-LIM is that it assumes genes act through independent gene-specific molecular phenotypes whose nonlinear interactions determine fitness. When this assumption holds, the model yields accurate predictions and interpretable effective phenotypes. Conversely, failure reveals that a low-dimensional model is insufficient. Applied to deep mutational scanning of metabolic pathways, protein-protein interactions, and yeast environmental adaptation, D-LIM achieves state-ofthe-art predictive accuracy. The inferred phenotype-fitness landscapes reveal whether epistatic interactions can be captured by a low-dimensional continuous model and identify potential trade-offs. Moreover, D-LIM estimates mutational effects on the effective phenotypes, enabling weak extrapolation beyond the training domain. D-LIM demonstrates how simple structure constraints in a neural network can help inference and hypothesis generation in biology.</div
Los micro-trabajadores detrás de la inteligencia artificial: Explorando nuevos sujetos digitales y sus precariedades en el mundo laboral
No full textInternational audienceIn digital environments, value production involves not only computer developers and engineers, but a broader range of digital subjects – from users to data workers – whose contributions are often occluded from view. We break down their digital labour into its different forms: classification, monetisation and automation in the case of users; and preparation, verification and impersonation in the case of data workers. Far from a simple succession of predefined mechanical tasks, we show that all these forms of work are complex human activities that harness knowledge, skills, personal commitments, moral judgements, emotional elements and bodily dimensions. When we open the black box of AI, what emerges is a plurality of subjects who, through their digital interactions, reveal intimate aspects of their subjectivities and form an essential—though largely overlooked—part of the value chain that sustains this technology. Therefore, any critical reflection on the regulation of AI and its ethical and social implications must recognise the active role played by these digital subjects as co-producers of value and invisible protagonists of the ongoing technological transformation.En los entornos digitales, la producción de valor no solo involucra a desarrolladores e ingenieros informáticos, sino a una gama más amplia de sujetos digitales —desde usuarios hasta trabajadores de datos— cuyas contribuciones a menudo quedan ocultas a la vista. Descomponemos su trabajo digital en sus diferentes formas: clasificación, monetización y automatización en el caso de los usuarios; y preparación, verificación y suplantación en el caso de los trabajadores de datos. Lejos de ser una simple sucesión de tareas mecánicas predefinidas, demostramos que todas estas formas de trabajo son actividades humanas complejas que aprovechan conocimientos, habilidades, compromisos personales, juicios morales, elementos emocionales y dimensiones corporales. Al abrir la caja negra de la IA, lo que emerge es una pluralidad de sujetos que, a través de sus interacciones digitales, revelan aspectos íntimos de sus subjetividades y forman una parte esencial —aunque en gran medida ignorada— de la cadena de valor que sustenta esta tecnología. Por lo tanto, cualquier reflexión crítica sobre la regulación de la IA y sus implicaciones éticas y sociales debe reconocer el papel activo que desempeñan estos sujetos digitales como coproductores de valor y protagonistas invisibles de la transformación tecnológica
On the Possibility of Melting Water Ice During the Recent Past of Mars: Implications for the Formation of Gullies
No full textInternational audienceAbstract The formation of gullies on Mars has often been attributed to the melting of (sub)surface water ice. However, melting‐based hypotheses generally overlook key processes: (a) sublimation cooling by latent heat absorption, (b) the non‐stability of ice where melting conditions can be reached, and (c) the particular microclimates of gullied slopes. Using state‐of‐the‐art climate simulations, we reassess ice melting scenarios over the past 4 million years (obliquity 35°), beyond the estimated period of gully formation. We find that the melting of opaque water snow or ice at the surface of Mars is unlikely anywhere due to sublimation cooling, while (quasi‐) stable subsurface ice is typically too deep to reach melting temperatures. We propose an alternative mechanism in which seasonal frost sublimation destabilizes the regolith and brings the underlying water ice close to the surface, allowing rapid heating. Even under these optimal conditions, melting requires unrealistic assumptions. Ice containing a small amount of dust could melt via a solid‐state greenhouse effect, but both its possibility and frequency in Mars' recent past remain uncertain