Agenda

Département
Langue
Date
Thématique
2025

December

  • 09:00
    18:00

    The workshop highlights the careers and experiences of women in computer science research and aims to inspire female engineering students.
    Link to follow this workshop.

    En distanciel
  • 11:00
    12:00

    Hussein Kazemi (LaBRI)

    Title: Trajectory visibility at first sight

    Abstract:

    Let P be a simple polygon with n vertices, and let two moving entities q(t) and r(t) travel at constant (possibly distinct) speeds vq and vr along line-segment trajectories τq and τr inside P. We study the exact first-visibility time t∗= min t≥0: q(t) r(t) ⊆P, the earliest moment at which the segment joining q(t) and r(t) lies entirely within P.

    Prior work by Eades et al. focused on this question in the setting of a simple polygon. They gave a one-shot decision algorithm running in O(n) time. For a stationary entity and a moving one, they suggested a structure that, after O(n log n) pre-processing, answers the decision query in O(log n) time, requiring O(n) space. In addition, for moving entities, after preprocessing time of O(n log⁵ n), they construct a data structure with O(n^{3/4} log³ n) query time and O(n log⁵ n) space. Variants for polygonal domains with holes or when entities cross the boundary of P lie beyond our scope.

    In this work, we go beyond the decision to compute t* exactly under three models for a simple polygon P. When both trajectories are known in advance, we preprocess P in O(n) time and space and thereafter answer each query in O(log n) time. If one trajectory τr is fixed while τq is given as query, we build a structure in O(n log n) time and space that computes t∗in O(log² n) time per query. In a setting where the trajectories are not known in advance, we develop a randomized structure with O(n^{1+ε}) expected pre-processing time and O(n) space, achieving an O(√n polylog(n)) expected query time for any fixed ε > 0.

    https://algodist.labri.fr/index.php/Main/GT

    English
    LaBRI 178
  • 09:30
    12:00

    The agricultural sector is undergoing a triple crisis, not only social but also economical and ecological, which requires a change in the production model. For political institutions, the solution lies in a twin transition, using information and communication technologies (ICT) as a lever to make the sector more sustainable. However, there are many uncertainties regarding the consequences of such a strategy. In particular, from an environmental point of view, few studies assess the effects of large-scale deployment of digital agriculture technologies. Yet the harmful effects of ICT on the environment are well known, and an increase in digital equipment and infrastructure for agriculture could be a threat to the sector’s sustainability. This thesis introduces methods for assessing the environmental impacts of large-scale deployment of digital agricultural equipment and its dependence on infrastructure. The general framework adopts a prospective scenario-based approach and parametric and consequential modelling. A first contribution concerns the estimation of the impacts of digital technologies deployed on farms in a given agricultural context. This context is defined as a type of production and a distribution of farms of varying sizes. This method is applied to two case studies in mainland France: the identification and monitoring of oestrus in dairy cattle, and robots for mechanical weeding and automated sowing in large-scale cereal crops. The results reveal heterogeneous impacts, depending on the number of pieces of equipment deployed, their mass and their technological complexity. Scenarios involving the most advanced technologies and using a larger mass of equipment tend to have the highest impacts. A second contribution extends the case study dedicated to weeding robots and focuses on the impacts associated with their use of the mobile network. This method simulates the deployment of robots in a given area, based on a set of agricultural parcels and existing mobile network infrastructure, with each site supporting a maximum data volume. The scenarios considered vary according to the design of the robots, and the volumes of data sent by the robots differ greatly from one design to another. Based on a traffic simulation of data and its impact on the mobile network, we estimate the additional impact on existing infrastructure in terms of embodied carbon footprint of required network equipment, and that associated with their energy consumption. We also estimate the proportion of agricultural land that can be managed by the deployed robots. The results show that for intensive use of the network, the capacity of the current network greatly limits the number of robots that can be deployed, even when the capacity of existing sites is maximized. Furthermore, for the most intensive scenarios, the additional power consumption of the network is of the same order of magnitude as that of the robots themselves. Thus, for these scenarios, only a small fraction of the parcels can be managed without adding sites, and increasing this fraction would imply a significant increase in the overall footprint of the mobile network in question. This manuscript provides an initial overview of the consequential carbon footprint associated with the deployment of digital equipment on farms and their use of the mobile network.

    Amphi LaBRI
  • 14:00
    18:00

    Magnetic resonance imaging is now an essential tool for studying the brain and diagnosing many neurological diseases. It allows the brain's structure and abnormalities to be observed non-invasively. Among the various acquisition sequences, the FLAIR sequence plays an essential role in detecting brain lesions. These images are widely used in clinical practice, particularly for conditions such as multiple sclerosis, brain tumors, and strokes. However, their automatic analysis remains difficult due to the low contrast between tissues, the limited signal-to-noise ratio, and the variability between imaging devices. In neuroimaging, quantitative analysis of the brain increasingly relies on automatic segmentation tools capable of identifying anatomical structures and pathological regions.
    Most of these methods were designed for T1-weighted images, which are considered the gold standard for anatomical segmentation. However, this sequence is not systematically acquired in clinical protocols, which limits the use of existing tools. In many cases, the FLAIR sequence is the only one available, but conventional approaches, designed for T1w, fail to provide reliable results. This thesis addresses this issue and explores the possibility of performing fine segmentation of the brain using FLAIR images alone. The objective is to develop robust methods capable of extracting relevant anatomical and pathological information from this sequence, while adapting to acquisition variability and the presence of lesions. The work carried out has led to the development of tools specifically adapted to FLAIR, designed to improve the performance of preprocessing and segmentation steps in a realistic clinical setting.
    The results obtained show that it is possible to obtain fine and consistent brain segmentations from FLAIR alone, paving the way for the exploitation of vast clinical databases that have been underutilized until now. These approaches also demonstrate the value of designing specialized models for each modality rather than generic methods that are independent of the sequence. Ultimately, the tools developed could be integrated into the volBrain online platform (https://volbrain.net) to facilitate their dissemination to the scientific and clinical community.
     

    Amphi LaBRI
  • 14:00
    18:00

    Systems biology studies the interactions and behavior of the components of biological entities, such as molecules, cells, organs, and organisms. This discipline emphasizes the central role of complex interactions within biological systems, as opposed to studying their components in isolation.
    The regulation of gene expression has important implications for both fundamental biology and numerous biomedical applications. The study of gene expression in systems biology relies on mathematical and computational modeling of experimental data. These have seen significant progress over the last decade with the advent of single-cell RNA sequencing (scRNA-seq), which
    allows gene expression to be measured with unprecedented resolution. A first step in modeling and understanding gene expression is to infer the interactions through which genes might influence each other. Currently, statistical and deep learning methods are used, in parallel with bioinformatic analyses, to deduce plausible interactions from scRNA-seq and other data sources. The set of inferred interactions then constitutes a gene regulatory network (GRN), a static model of the influences between genes. GRNs provide valuable information, but their description does not explain the regulatory mechanisms behind the observed biological phenomena. Dynamic models aim to fill this gap with a mechanistic view capable of explaining and reproducing the variations observed in the data. Boolean networks (BNs) are a special class of dynamic systems commonly used to model gene expression regulation. In these models, the activity of biological entities is represented as either active or inactive. This binary representation allows us to reason about causal relationships between entities without having to estimate kinetic parameters or regulatory thresholds. Despite recent advances in this field, many questions regarding the modeling of scRNA data using BNs remain open. In my thesis, I am working on two main areas concerning BNs and scRNA-seq data.
    The first axis consists of linking scRNA-seq data, which is quantitative in nature, to the qualitative states of BNs. To this end, my first contribution is the new scBoolSeq method, which provides a qualitative interpretation of experimental scRNA-seq data and also allows the generation of synthetic scRNA-seq data that takes into account the Boolean states of gene activation. scBoolSeq has already been used to compare inference methods using the aforementioned synthetic scRNA-seq data reflecting Boolean dynamics. The second approach consists of applying the model training and evaluation methodology used in statistical learning to BN inference. This is often done manually and only takes isolated experiments into account. My proposal is to exploit several experimental conditions as follows: By inferring Boolean networks on “training” experimental conditions and measuring their predictive power on validation data, it is possible to evaluate and subsequently optimize their predictive power. Thus, my research represents a new direction that combines formal methods with statistical learning to improve the modeling of scRNA-seq data using Boolean networks.

    Amphi LaBRI
  • 09:30
    12:00

    Magnetic resonance imaging is now an essential tool for studying the brain and diagnosing many neurological diseases. It allows the brain's structure and abnormalities to be observed non-invasively. Among the various acquisition sequences, the FLAIR sequence plays an essential role in detecting brain lesions. These images are widely used in clinical practice, particularly for conditions such as multiple sclerosis, brain tumors, and strokes. However, their automatic analysis remains difficult due to the low contrast between tissues, the limited signal-to-noise ratio, and the variability between imaging devices. In neuroimaging, quantitative analysis of the brain increasingly relies on automatic segmentation tools capable of identifying anatomical structures and pathological regions.
    Most of these methods were designed for T1-weighted images, which are considered the gold standard for anatomical segmentation. However, this sequence is not systematically acquired in clinical protocols, which limits the use of existing tools. In many cases, the FLAIR sequence is the only one available, but conventional approaches, designed for T1w, fail to provide reliable results. This thesis addresses this issue and explores the possibility of performing fine segmentation of the brain using FLAIR images alone. The objective is to develop robust methods capable of extracting relevant anatomical and pathological information from this sequence, while adapting to acquisition variability and the presence of lesions. The work carried out has led to the development of tools specifically adapted to FLAIR, designed to improve the performance of preprocessing and segmentation steps in a realistic clinical setting.
    The results obtained show that it is possible to obtain fine and consistent brain segmentations from FLAIR alone, paving the way for the exploitation of vast clinical databases that have been underutilized until now. These approaches also demonstrate the value of designing specialized models for each modality rather than generic methods that are independent of the sequence. Ultimately, the tools developed could be integrated into the volBrain online platform (https://volbrain.net) to facilitate their dissemination to the scientific and clinical community.

     

    Amphi LaBRI
  • 09:30
    12:00

    Oussama Laaroussi will defend his thesis on December 19, 2025, at 9:30 a.m. in the LaBRI lecture hall.
    The title of his thesis is: “An adaptive recommendation system based on incremental classification in an e-learning environment.”

     

    Amphi LaBRI