Work trips.

Whenever possible, we prefer using train rather than plane for national and European travels. Most of us continue to submit papers to international conferences outside of Europe but if a paper gets accepted into such conference, we priorize sending the first author (PhD student). Outside of conferences, for national events (seminars, PhD juries, etc.), videoconference is increasingly used, which helps to reduce the overall carbon footprint of the community.

Utilisation of computing resources.

The discontinuation of Igrida services and the transition towards Grid'5000 and Jean Zay has reduced our access to easily available computation resources. It adds friction to making experiments, but as a positive effect on energy consumption, as we are now using national infrastructures that benefit from even better sharing between users than Igrida (which was already heavily used).

7.1.1 HIPAR

• Name:
  Hierarchical Interpretable Pattern-aided Regression
• Keywords:
  Regression, Pattern extraction
• Functional Description:
  Given a (tabular) dataset with categorical and numerical attributes, HIPAR is a Python library that can extract accurate hybrid rules that offer a trade-off between (a) interpretability, (b) accuracy, and (c) data coverage.
• URL:
  https://­gitlab.­inria.­fr/­lgalarra/­hipar
• Contact:
  Luis Galarraga Del Prado

7.1.2 Dexteris

• Keywords:
  Data Exploration, Querying, Interactive method, JSon
• Functional Description:

  Dexteris is a low-code tool for data exploration and transformation. It works as an interactive data-oriented query builder with JSONiq as the target query language. It uses JSON as the pivot data format but it can read from and write to a few other formats: text, CSV, and RDF/Turtle (to be extended to other formats).

  Dexteris is very expressive as JSONiq is Turing-complete, and supports a varied set of data processing features: - reading JSON files, and CSV as JSON (one object per row, one field per column), - string processing (split, replace, match, ...), - arithmetics, comparison, and logics, - accessing and creating JSON data structures, i- terations, grouping, filtering, aggregates and ordering (FLWOR operators), - local function definitions.

  The built JSONiq programs are high-level, declarative, and concise. Under-progress results are given at every step so that users can keep focused on their data and on the transformations they want to apply.

• URL:
  http://­www.­irisa.­fr/­LIS/­ferre/­dexteris/
• Publication:
  hal-04186117v1
• Contact:
  Sebastien Ferre

7.1.3 skm

• Name:
  scikit-mine
• Keywords:
  Artificial intelligence, Data mining, Pattern discovery, Sequential patterns
• Functional Description:
  The library offers several algorithms for extracting a reasonable-sized set of patterns for different types of data (itemsets, sequences, graphs).
• URL:
  https://­scikit-mine.­github.­io/­scikit-mine/
• Contact:
  Peggy Cellier

8.1.1 Pattern Mining

Participants: Tassadit Bouadi, Sébastien Ferré, Luis Galárraga.

Remark about the “Participants” boxes: we compiled syntactically the list of co-authors of the papers that make the “New Results” of the year, for each subsection. It obviously does not mean that other members of the team do not work on the topics listed, the correct meaning is that they did not have a publication on that topic this year.

8.1.2 Graph-FCA

Participants: Tassadit Bouadi, Peggy Cellier, Sébastien Ferré, Luis Galárraga.

Some of previously presented documents also contribute to this research domain: 41 .

8.1.3 Semantic Web

Participants: Tassadit Bouadi, Peggy Cellier, Sébastien Ferré, Luis Galárraga.

Some of previously presented documents also contribute to this research domain: 41 .

A PAC-Bayesian Bound on a Risk Measure for Fair Learning 37.

We study the Conditional Value at Risk (CVaR), a risk measure defined as a combination of subgroup risks that focuses solely on the highest risks. In fair learning, a subgroup is defined by the set of individuals sharing the same value of the sensitive attribute under consideration. CVaR is thus valuable as it concentrates on the values of the sensitive attribute associated with high risk. In this work, we derive a PAC-Bayesian generalization bound for CVaR that involves not only a distribution over the set of hypotheses (as is common in PAC-Bayes), but also a distribution over the set of sensitive attribute values, allowing control over the importance of each subgroup contributing to the CVaR.

Explaining a Black Box Without a Black Box 21.

Counterfactual explanation methods are popular approaches for explaining machine learning algorithms. These explanations encode the necessary modifications in a target document to change a classifier's prediction. Most of these methods find such explanations by iteratively perturbing the target document until it is classified differently by the black box. We identify two main families of counterfactual approaches in the literature: (a) "transparent" methods that perturb the target by adding, removing, or replacing words, and (b) "opaque" techniques that project the target document into a non-interpretable latent space where the perturbation is then performed. This contribution presents a comparative study of the performance of these two families of methods on three standard natural language processing tasks. Our results show that for applications such as fake news detection or sentiment analysis, opaque counterfactual approaches may introduce an additional level of complexity without significant improvement.

Uniform Generalization Bounds on Data-Dependent Hypothesis Sets via PAC-Bayesian Theory on Random Sets 43.

We propose data-dependent uniform generalization bounds by approaching the problem from a PAC-Bayesian perspective. We first apply the PAC-Bayesian framework on `random sets' in a rigorous way, where the training algorithm is assumed to output a data-dependent hypothesis set after observing the training data. This approach allows us to prove data-dependent bounds, which can be applicable in numerous contexts. To highlight the power of our approach, we consider two main applications. First, we propose a PAC-Bayesian formulation of the recently developed fractal-dimension-based generalization bounds. The derived results are shown to be tighter and they unify the existing results around one simple proof technique. Second, we prove uniform bounds over the trajectories of continuous Langevin dynamics and stochastic gradient Langevin dynamics. These results provide novel information about the generalization properties of noisy algorithms.

A PAC-Bayesian Link Between Generalisation and Flat Minima 47.

Modern machine learning usually involves predictors in the overparametrised setting (number of trained parameters greater than dataset size), and their training yield not only good performances on training data, but also good generalisation capacity. This phenomenon challenges many theoretical results, and remains an open problem. To reach a better understanding, we provide novel generalisation bounds involving gradient terms. To do so, we combine the PAC-Bayes toolbox with Poincaré and Log-Sobolev inequalities, avoiding an explicit dependency on dimension of the predictor space. Our results highlight the positive influence of flat minima (being minima with a neighbourhood nearly minimising the learning problem as well) on generalisation performances, involving directly the benefits of the optimisation phase.

Shaping Up SHAP: Enhancing Stability through Layer-Wise Neighbor Selection 29.

Machine learning techniques, such as deep learning and ensemble methods, are widely used in various domains due to their ability to handle complex real-world tasks. However, their black-box nature has raised multiple concerns about the fairness, trustworthiness, and transparency of computerassisted decision-making. This has led to the emergence of local post-hoc explainability methods, which offer explanations for individual decisions made by black-box algorithms. Among these methods, Kernel SHAP is widely used due to its model-agnostic nature and its well-founded theoretical framework. Despite these strengths, Kernel SHAP suffers from high instability: different executions of the method with the same inputs can lead to significantly different explanations, which diminishes the utility of post-hoc explainability. The contribution of this work is twofold. On the one hand, we show that Kernel SHAP's instability is caused by its stochastic neighbor selection procedure, which we adapt to achieve full stability without compromising explanation fidelity. On the other hand, we show that by restricting the neighbors generation to perturbations of size 1-which we call the coalitions of Layer 1-we obtain a novel feature attribution method that is fully stable, computationally efficient, and still meaningful.

Synthetic Data: Generate Avatar Data on Demand 30.

Anonymization is crucial for the sharing of personal data in a privacy-aware manner yet it is a complex task that requires to set up a trade-off between the robustness of anonymization (i.e., the privacy level provided) and the quality of the analysis that can be expected from anonymized data (i.e., the resulting utility). Synthetic data has emerged as a promising solution to overcome the limits of classical anonymization methods while achieving similar statistical properties to the original data. Avatar-based approaches are a specific type of synthetic data generation that rely on local stochastic simulation modeling to generate an avatar for each original record. While these approaches have been used in healthcare, their attack surface is not well documented and understood. In this contribution, we provide an extensive assessment of such approaches and comparing them against other data synthesis methods. We also propose an improvement based on conditional sampling in the latent space, which allows synthetic data to be generated on demand (i.e., of arbitrary size). Our empirical analysis shows that avatar-generated data are subject to the same utility and privacy trade-off as other data synthesis methods with a privacy risk more important on the edge data, which correspond to records that have the fewest alter egos in the original data.

A Theoretically Grounded Extension of Universal Attacks from the Attacker's Viewpoint 32.

We extend universal attacks by jointly learning a set of perturbations to choose from to maximize the chance of attacking deep neural network models. Specifically, we embrace the attacker's perspective and introduce a theoretical bound quantifying how much the universal perturbations are able to fool a given model on unseen examples. An extension to assert the transferability of universal attacks is also provided. To learn such perturbations, we devise an algorithmic solution with convergence guarantees under Lipschitz continuity assumptions. Moreover, we demonstrate how it can improve the performance of state-of-the-art gradient-based universal perturbation. As evidenced by our experiments, these novel universal perturbations result in more interpretable, diverse, and transferable attacks.

Tighter Generalisation Bounds via Interpolation 49.

This contribution contains a recipe for deriving new PAC-Bayes generalisation bounds based on the $(f,\Gamma )$-divergence, and, in addition, presents PAC-Bayes generalisation bounds where we interpolate between a series of probability divergences (including but not limited to KL, Wasserstein, and total variation), making the best out of many worlds depending on the posterior distributions properties. We explore the tightness of these bounds and connect them to earlier results from statistical learning, which are specific cases. We also instantiate our bounds as training objectives, yielding non-trivial guarantees and practical performances.

Leveraging PAC-Bayes Theory and Gibbs Distributions for Generalization Bounds with Complexity Measures 34.

In statistical learning theory, a generalization bound usually involves a complexity measure imposed by the considered theoretical framework. This limits the scope of such bounds, as other forms of capacity measures or regularizations are used in algorithms. In this contribution, we leverage the framework of disintegrated PAC-Bayes bounds to derive a general generalization bound instantiable with arbitrary complexity measures. One trick to prove such a result involves considering a commonly used family of distributions: the Gibbs distributions. Our bound stands in probability jointly over the hypothesis and the learning sample, which allows the complexity to be adapted to the generalization gap as it can be customized to fit both the hypothesis class and the task.

8.3.1 Computer Vision and Robotics

Participants: Élisa Fromont, Elodie Germani, Yasmine Hachani.

8.3.2 Agriculture

Participants: Élisa Fromont, Luis Galárraga, Olivier Gauriau, Yasmine Hachani, Alexandre Termier.

Some of previously presented documents also contribute to this research domain: 50 .

8.3.3 Machine Learning on Sequences

Participants: Élisa Fromont, Luis Galárraga, Antonin Voyez.

10.1.1 Visits of international scientists

10.2.1 Horizon Europe

Several unsuccessul submissions at the European level :

• Elisa Fromont participated in the submission of the PROSPER (Pioneering Responsible disclosure of multimodal healthcare data Openly by data Synthesis and the Prototyping of applications for hearing and balance Evaluation and Restoration) project: call HORIZON-JU-IHI-2023-05 (resubmition in 2025)

• Romaric Gaudel participated in the submission of the DoCA-AI (Intelligent on-demand Detection and Simulation of False Data Injection Attacks in Global Maritime Surveillance) project. call : European Defence Fund-2023-LS-RA-SMERO.

  This project aims to explore novel on-demand interpretable cyber-threat detection and simulation capabilities dedicated to the of the global maritime surveillance and awareness (global MSA). The project focuses on detecting False AIS tracks by leveraging explainable AI methods and combining multi-sources of data.

  The main contribution of LACODAM's members is on the developpment of XAI models and methods relevant dedicated to time-series and therefore relevant for AIS data.

10.2.2 H2020 projects

Luis Galarraga Del Prado , Alexandre Termier and Elisa Fromont were part of the TAILOR Network within Inria. TAILOR was one of four “Network of Excellence” working on aspects of trustworthy AI funded under the H2020-ICT-48-2020 call.

10.3.1 ANR

• FAbLe: Framework for Automatic Interpretability in Machine Learning

  Participants: L. Galárraga (holder), C. Largouët

  Participants: Luis Galárraga (holder), Christine Largouët, Julien Delaunay, Julianne Guerbette.

  How can we fully automatically choose the best explanation for a given use case in classification?. Answering this question is the raison d’être of the JCJC ANR project FAbLe. By “best explanation” we mean an explanation that is both understandable by humans and faithful among a universe of possible explanations. We focus on local explanations, i.e., when we want to explain the answer of a black-box model for a given use case, which we call the “target instance”. We argue that the choice of the best explanation depends on the (i) data, namely the model, the explanation technique and the target instance, etc., and (ii) the recipients of the explanations. Hence our research is focused on two main questions: “What makes an explanation suitable (interpretable and faithful) for a particular instance and model?” and “What is the effect of the different AI-based explanation techniques and visual representations on users' comprehension and trust?”. Answering these questions will help us understand and automate the selection of a particular explanation style based on the use case. Our ultimate goal is to produce a suite of algorithms that will compute suitable explanations for ML algorithms based on our insights of what is interpretable. User studies on different explanation settings (methods and visual representations) will allow us to characterize the features of explanations that make them acceptable (i.e., understandable and trustworthy) by users.

• SmartFCA: A Smart Tool for Analyzing Complex Data with Formal Concept Analysis

  Participants: Sébastien Ferré, Peggy Cellier, Frederic Lang.

  Period: 01/01/2022 – 31/12/2025

  Budget: 143k€ (Univ Rennes)

  Formal Concept Analysis (FCA) is a mathematical framework based on lattice theory and aimed at data analysis and classification. FCA, which is closely related to pattern mining in knowledge discovery (KD), can be used for data mining purposes in many application domains, e.g. life sciences and linked data. Moreover, FCA is human-centered and provides means for visualization and interaction with data and patterns. Actually it is now possible to deal with complex data such as intervals, sequences, trajectories, trees, and graphs. Research in FCA is dynamic, but there is still room for extensions of the original formalism. Many theoretical and practical challenges remain. Actually there does not exist any consensual platform offering the necessary components for analyzing real-life data. This is precisely the objective of the SmartFCA project to develop the theory and practice of FCA and its extensions, to make the related components inter-operable, and to implement a usable and consensual platform offering the necessary services and workflows for KD.

  In particular, for satisfying in the best way the needs of experts in many application domains, SmartFCa will offer a “Knowledge as a Service” (KaaS) component for making domain knowledge operable and reusable on demand.

• MeKaNo: Search the Web with Things

  Participants: Sébastien Ferré, Peggy Cellier, Luis Galárraga, Aurélien Lamercerie.

  Period: 01/10/2022 – 29/09/2026

  Budget: 143k€ (Univ Rennes)

  In MeKaNo, we aim to search the web with things, in order to get more accurate results over a wide diversity of sources. Traditional web search engines search the web with strings. However, keyword search often returns many irrelevant documents, pushing users to refine their keyword list following a trial-and-error process. To overcome such limitations, major companies allowed searching for things, not strings. Asking for the age of “James Cameron” to your vocal assistant, it locates in a Knowledge Graph (KG) a Person matching “James Cameron” where a property “age” is set to 66 years, i.e. the Thing “James Cameron”. If searching for Things is a tremendous progress and delivers exact answers, the search is done over a Knowledge Graph and not on the Web. Consequently, there may exist many answers on the web that are not part of the knowledge graph.

  To summarize, searching with strings over the web offers diversity at the expense of noise. Searching for Things delivers exact answers, but we lose diversity. In MeKaNo, we aim at searching the web with Things to get diversity and avoid noisy results. To search the web with Things, we face three main scientific challenges:

  1. Users are used to search with keywords. Transforming a keyword query into a mixed query that first searches over a KG then into the web is difficult, especially, for complex queries.
  2. As with traditional web searches, users expect to obtain ranked results in a snap. Combining KG search and Web search while preserving performances is highly challenging and requires a new kind of search engine.
  3. Improving the connection between the web of microdata and Knowledge Graphs requires entity matching at large scale for microdata entities and KG entities.

11.1.1 Scientific events: organisation

11.1.2 Scientific events: selection

11.1.3 Journal

11.1.4 Invited talks

• Luis Galárraga gave a course on Explicabilité et “Machine Learning” Supervisé (Explainability and Supervised Machine Learning) at the MDD'24 (Masse de Données) Summer School organized by the French Community on Data Management on June 2024.
• Paul Viallard was invited for presenting his paper “Tighter Generalisation Bounds via Interpolation” to the Journée MAS 2024.
• Peggy Cellier and Sebastien Ferré were invited to give a talk on “Étude de l'utilisation des modèles de langages pour l'interrogation en langue naturelle des graphes de connaissances” at a series of seminar on Geo-Historical Knowledge Graphs, organized by GDR MAGIS, on 7 November 2024.
• Alexandre Termier was invited to give a talk on “Introduction to Generative AI” at the scientific committee of the PHASE department of Inrae, on the 23rd of May 2024. He also gave this talk at an internal seminar of Inrae Centre Val de Loire on the 27th of September 2024. Moreover, he presented the WAIT4 PEPR project at the annual meeting of the DigitAg project on the 23rd of April 2024.
• Elisa Fromont was Keynote (in French) on « Conférence sur l'I.A.", Auditorium de l'ESCC, Maen Roch, France. The 4/04/2024. She was panellist (in French) at the Techno-conférence on Generative AI organisée par le pôle de compétitivité I&R on the 10/10/2024. She was keynote speaker (in French) at Forum campus psy 2024 : Désir d’intelligence artificielle ? on "L'IA et le traitement de la parole", IFSI Rennes on the 12/10/2024. She was invited speaker (in French) at Les Territoriales de Bretagne on «IA et société : là ou le bât blesse (et ce qu’on peut y faire) », Saint Brieuc on the 3/12/2024 and speaker and Panellist (in French) at the Séminaire DIGIT-BIO on "Data augmentation with style transfer", Lyon.
• Christine Largouët was invited by the Consulate General of France in Chicago to give a talk during the French American Innovation Day, Digital Ag & AI, on "Artificial Intelligence and Agriculture", 06/02/2024. She was keynote speaker (in French) at the seminar Ermerg'IN LiPh4SAS, INRAE on "Intelligence Artificielle pour le pilotage des élevages", 16/05/2024.

11.1.5 Leadership within the scientific community

• Elisa Fromont is in the steering committee of the International Conference IDA and of the French Conference on machine learning (CAp).

11.1.6 Scientific expertise

• Alexandre Termier reviewed a project for the Knowledge Foundation (Sweden), an application for an associate professor position at the University of Haifa (Israel), and an associate team proposal for Inria.
• Elisa Fromont As a member of the CSV ("Comité de Sélection et de Validation") of Images & Réseaux (pôle de compétitivité) she reviewed industrial projects all year long. She also reviewed project as a member of the scientific council of the MathNum dept of Inrae.
• Peggy Cellier was member of an associate professor recruitement committee in La Réunion (Université de La Réunion).
• Christine Largouët took part in the evaluation process of ANR-AAPG 2024 as a Scientific Expert for the scientific panel CE23 - Artificial intelligence and data science.

11.1.7 Research administration

• Peggy Cellier is in charge of the Phd students of the IRISA lab (commission personnel each month, etc). She was elected at the "Conseil de coposante" of the Computer Science departement of INSA Rennes until the end of 2024. She is also a member of "Conseil de l'école doctorale MATISSE".
• Romaric Gaudel: Romaric Gaudel is elected member of IRISA's board since dec. 2024.
• Elisa Fromont is the head of the D7 dpt of IRISA. She is elected at the scientific council of Université de Rennes and as such, she is also a member of the academic council (CAC) of the University and a member of the HDR Committee of the University. She is a member of the executive board of the CominLabs LABEX for the "Data, IA and Robotics" scientific axis.

11.2.1 Teaching

Apart from Luis Galárraga and Paul Viallard (research scientists), and Gaelle Tworkowski (administrative assistant), each permanent member of the project-team LACODAM is also faculty members and is actively involved in computer science teaching programs in ISTIC, IUT of Lannion, INSA, or Agrocampus-Ouest. Besides these usual teachings LACODAM is responsible of some teaching tracks and of some courses.

11.2.2 Supervision

PhD. Students

• (defended in 2024) Olivier Gauriau, (Inria, DigitAg, Acta Toulouse) 2021-2024; supervisors: Luis Galárraga , François Brun, Alexandre Termier and David Makowski; title: Numerical Rule Mining for the Prediction of the Dynamics of Crop Diseases, ED Matisse.
• (defended in 2024) Elodie Germani, 2021-2024; supervisors: Elisa Fromont and Camille Maumet (EMPENN); title: on representation learning for more robust FMRI data analysis, ED Matisse.
• Gwladys Kelodjou, 2022-2025; supervisors: Véronique Masson , Laurence Rozé , Alexandre Termier ; title: Beyond the oracle: stabilizing the interpretability of machine learning algorithms, ED Matisse.
• Charbel Kindji, 2022-2025; supervisors: Elisa Fromont and Tanguy Urvoy (OrangeLabs); title: Architectures connexionnistes pour la génération de données tabulaires, ED Matisse.
• Lucie Lepetit, 2022-2025; supervisors: Peggy Cellier , Bruno Crémilleux and Alexandre Termier ; title: Data mining methods for discovering behaviors related to animal well-being in precision farming data, ED Matisse.
• Pierre Maurand, 2022-2025; supervisors: Tassadit Bouadi , Peggy Cellier , Bruno Crémilleux (GREYC) and Alexandre Termier ; title: Tell me your preferences and I will show you what you are interested in, ED Matisse.
• Vanessa Fokou, 2022-2025; supervisors: Florence Le Ber and Xavier Dolques (Univ. Strasbourg), Sebastien Ferre , Peggy Cellier ; title: Comparison and cooperation of different Formal Concept Analysis approaches for relational data, Univ. Strasbourg.
• Isseinie Sinouvassane, 2023-2026; supervisors: Luis Galarraga , Alexandre Termier ; title: How-Provenance Polynomials for Efficient and Greener Rule Mining, ED Matisse (financed by an ENS scholarship).
• Yasmine Hachani, 2023-2026; supervisors: Patrick Bouthémy (SAIRPICO), Elisa Fromont ; title: Analyse par apprentissage profond de la dynamique du développement précoce des embryons bovins à partir de vidéomicroscopie, ED Matisse.
• Paul Sevellec, 2023-2026; supervisors: Matteo Sammarco (Stellantis), Elisa Fromont , Romaric Gaudel , Laurence Rozé ; title: Explications de séries temporelles multivariées par contrefactuels, ED Matisse.
• Dimitri Lerévérend, 2023-2026; supervisors: Davide Frey (WIDE), Romaric Gaudel ; title: Privacy Preserving Decentralized Through Model Fragmentation, ED Matisse.
• Niels Cobat, 2024-2027; supervisors: Romaric Gaudel , Damien Hardy (PACAP) ; title: Analyse et optimisation des fichiers d'impression 3D à l'aide de méthodes d'apprentissage automatique, ED Matisse.
• Ismail Bachchar, 2024-2027; supervisors: Tassadit Bouadi , Thomas Guyet (AISTROSIGHT) ; title: Conception d’explications individuelles locales robustes, ED InfoMaths.
• Sacha Germain, 2024-2027; supervisors: Christine Largouët , Laurence Rozé , Tassadit Bouadi ; title: Detection et explications du comportement individuel et collectif des individus au sein d’un groupe pour estimer leur bien-être, ED Matisse.
• Manuel Nkegoum, 2024-2027; supervisors: Elisa Fromont , Sébastien Lefèvre (OBELIX) ; title: Object detection from few multispectral examples, ED Matisse.
• Youwan Mahé, 2024-2027; supervisors: Elisa Fromont , Elise Bannier (EMPENN) ; title: Détection et segmentation d'anomalies pour la caractérisation de la récuperation post AVC, ED Matisse.
• Nouha Karaouli, 2024-2027; supervisors: Elisa Fromont , Denis Coquenet (SHADOC) ; title: Incremental Deep Learning for Embedded Systems, ED Matisse.

Internships

• Hind Atbir (M2); supervisor: Paul Viallard , Farah Cherfaoui (Université de Saint-Etienne), Emilie Morvant (Université de Saint-Etienne), Guillaume Metzler (Université Lumière Lyon 2); title: PAC-Bayesian Fair Learning
• Niels Cobat (M2); supervisor: Romaric Gaudel , Damien Hardy (PACAP); title: Estimation précise du temps d'impression 3D par machine learning
• Sacha Germain (M2); supervisor: Christine Largouët (with Sébastien Picault, INRAE) ; title: Prediction of respiratory diseases occurrence in connected cattle farms with machine learning methods
• Loane Portier (L3); supervisor: Luis Galarraga Del Prado ; title: Multimodal Explainable Knowledge Graph Embeddings
• Amina Lahlah (L3); supervisor: Laurence Rozé ,Véronique Masson ; title: Tester et compléter une nouvelle version du logiciel Leftist
• Romain Eisenschmidt (L3); supervisor: Elisa Fromont ; title: Dépollution de la science
• Enzo Ouattara (L3); supervisor: Laurence Rozé ; title: Analyse et correction des données de recensement des marins français : stratégies d'évaluation et construction d'un rapport interactif

Engineer

• Julianne Guerbette, 2024 ; supervisor: Luis Galárraga ; Project: Different tasks in eXplainable AI. Julianne worked mainly in the put in shape of the AMIE rule mining system including a refactoring of its codebase, the development of a server backend for the storage (database) layer, and the development and evaluation of lightweight estimations of the rule metrics for the sake of speed-up and scalability.
• Frederic Lang, 2024-2025; supervisors: Sebastien Ferré , Peggy Cellier ; project: SmartFCA. Frédéric worked on the SmartFCA platform, and on Graph-FCA. He developed an OCaml version of part of the framework, and then used that to lift our Graph-FCA tool as a SmartFCA component that can be integrated into the platform.

Apprentice

• Loane Portier, 2024-2026; supervisor: Elisa Fromont ; Title: Post-hoc Explanations for Federated Learning Systems

11.2.3 PHD & HDR Juries

• Peggy Cellier was a member of the following PhD juries in 2024: Thomas Georges, 23/01 Univ. Montpellier (PhD, reviewer); Etienne Lehembre, 06/09 Univ. de Caen (PhD, examinatrice); Sarah Theroine, 20/12 Univ. Bourgogne (PhD, examinatrice).
• Elisa Fromont (14, 50% local): Rehan Juhboo, 10/01 Saint-Etienne (reviewer); Léonard Tschora, 17/01 Lyon (committee member, president); Lorenzo Perini 31/01 & 28/03, Leuven, Be (reviewer), Zoubida Ameur, 24/05, Rennes (president); Elodie Germani, Rennes, 16/09 (co-supervisor); Lincen Yang, Leiden, 20/09, NL (reviewer); Reda Marzouk, 17/10, Nantes; Malik Kazi Aoual, 20/11, Paris; Florent Imbert, 22/11, Rennes; Hasnaa Ouadoudi Belabzioui, 20/12, Rennes. HDR Rémi Emonet, 05/02, Saint-Etienne (president); Camille Maumet, 8/02, Rennes (president); Tristan Allard, 10/04, Rennes (president); Mathieu Lefort, 13/12, Lyon (committee member).
• Sebastien Ferre was a member of the following PhD juries in 2024: Sarah Ghidalia, 10/01 Univ. Bourgogne (rapporteur).
• Alexandre Termier was a member of the following PhD juries in 2024: Romain Brisse, 15/02 CentraleSupelec Rennes (examiner, president); Maxime Fuccellaro, 25/11 Univ. Bordeaux (reviewer); Cedric Gernigon, 18/12 Univ. Rennes (examiner)
• Romaric Gaudel was a member of the following jury: Ludovic DELANDE, 25/09 University of Toulouse (PhD, reviewer);

11.2.4 Doctoral advisory comitee (CSID)

• Peggy Cellier was a member of the mid-term evaluation committee of Oumaima El Khettari (Université de Nantes); Clémence Sebe (Université Paris Saclay); Randa Bendjeddou (Université de Lyon 2); Yacine Mokhtari (IMT Atlantique Brest) and Jules Berry (INSA Rennes, IRMAR).
• Luis Galárraga was a member of the mid-term evaluation committee of Ataollah Kamal (INSA Lyon) and Sacha Corbugy (Université de Namur), and Owen Le Godinec (University of Rennes)
• Romaric Gaudel was a member of the mid-term evaluation juries of Émile Binot.
• Elisa Fromont was a member of the mid-term evaluation juries of Ewan MOREL-CORLU (Rennes) 16/09/2024; Bruno Michelot (Lyon) 05/09/24; Loïc Eyango (Rennes) 14/06/2024; Irina Proskurina (Lyon) 2/05/2024; Ricky Walsh (Rennes) 3/04/2024.
• Laurence Rozé was a member of the mid-term evaluation juries of Nolwenn Pinczon du sel.
• Christine Largouët was a member of the mid-term evaluation juries of Maryem BEN SALEM (Université de Rennes) 21/06/2024.

11.3.1 Specific official responsibilities in science outreach structures

• Tassadit Bouadi is co-organizer of the project L Codent, L Créent Rennes, since 2018.
• Elisa Fromont is co-organizer of the project J'peux pas j'ai informatique for secondary school teacher, since 2022. She co-oragnized a "JFMI" day for (100) secondary school female students in December.

11.3.2 Productions (articles, videos, podcasts, serious games, ...)

• Luis Galárraga was co-author of the chapter Neurosymbolic Methods for Rule Mining of the book The Handbook on Neurosymbolic AI and Knowledge Graphs to be published by IOS Press series “Frontiers in Artificial Intelligence and Applications”

11.3.3 Participation in Live events

• Romaric Gaudel did the presentation "Comment intégrer l’intelligence artificielle et ses enjeux à l’enseignement de l’informatique ?" during "Les rendez-vous de l'informatique" at Rennes to, 11/04 and 12/04.
• Romaric Gaudel did the presentation "IA & Société, Là où le bât blesse…​ (et ce qu’on peut y faire)​" to Students at INSA Rennes, 15/10.
• Elisa Fromont on 21/02/2024, 15/03/2024, 5/04/2024 she gave a conference (in French) for secondary and high school students "Intelligence Artificielle de quoi parle-t-on ?". 2 conferences were done within the Chiche ! Inria program. On 19/03/24, she made a small videolecture for the 5 minutes of Lebesgue LABEX (in French) on "Les modèles de diffusion en apprentissage automatique". On 31/05/2024, 19/06/2024, 26/06/2024, she participated in training sessions in AI for Bachelor female Mathematicians involved in Math C pour L and (female-only) high school students.On 02/09/2024, she co-organized and gave an introductory conference at the AI summer school of the TIARe project on the "Societal Impacts of AI" in Rennes.
• Peggy Cellier gave a conference (in French) for Students at INSA "Intelligence Artificielle : quelques éléments de réflexion" in the context of "Concours d'éloquence 2024". She also gave a conference in french for the SAF (Service des Affaires Financières) team of Inria RBA about "Intelligence Artificielle".

11.3.4 Others science outreach relevant activities

• Luis Galárraga and Christine Largouët participated as experts guests at the episode “Dans le secret du fonctionnement des IA” of the scientific dissemination show Science en Question filmed on March 2024 and released on July 2024 by L'Esprit Sorcier TV

International journals

• 20 articleT.Tristan Allard, H.Hira Asghar, G.Gildas Avoine, C.Christophe Bobineau, P.Pierre Cauchois, E.Elisa Fromont, A.Anna Monreale, F.Francesca Naretto, R.Roberto Pellungrini, F.Francesca Pratesi, M.-C.Marie-Christine Rousset and A.Antonin Voyez. Analyzing and explaining privacy risks on time series data: ongoing work and challenges.SIGKDD explorations : newsletter of the Special Interest Group (SIG) on Knowledge Discovery & Data Mining2612024, 1-10In press. HALDOIback to text
• 21 articleJ.Julien Delaunay, L.Luis Galárraga and C.Christine Largouët. Expliquer une boîte noire sans boîte noire.Revue TAL : traitement automatique des langues643/20232024, 93-117HALback to text
• 22 articleO.Olivier Gauriau, L.Luis Galárraga, F.François Brun, A.Alexandre Termier, L.Loïc Davadan and F.François Joudelat. Comparing machine-learning models of different levels of complexity for crop protection: A look into the complexity-accuracy tradeoff.Smart Agricultural Technology7March 2024, 100380HALDOIback to text
• 23 articleS.Stefano Moia, H.-T.Hao-Ting Wang, A. S.Anibal S. Heinsfeld, D.Dorota Jarecka, Y. F.Yu Fang Yang, S.Stephan Heunis, M.Michele Svanera, B. D.Benjamin De Leener, A.Andrea Gondová, S.Sin Kim, A.Arshitha Basavaraj, J. M.Johanna M. M. Bayer, R. G.Roza G. Bayrak, P.-L.Pierre-Louis Bazin, I. P.Isil Poyraz Bilgin, S.Steffen Bollmann, D.Daniel Borek, V.Valentina Borghesani, T.Trang Cao, G.Gang Chen, A. d.Alejandro de La Vega, S.Sebastian Dresbach, P.Philipp Ehses, J.Jan Ernsting, I.Inês Esteves, O.Oscar Ferrante, K. G.Kelly G. Garner, R.Rémi Gau, E.Elodie Germani, T.Tara Ghafari, S. S.Satrajit S. Ghosh, S. E.Sarah E. Goodale, C. D.Cassandra D. Gould Van Praag, S.Samuel Guay, O. F.Omer Faruk Gulban, Y. O.Yaroslav O. Halchenko, M.Michael Hanke, P.Peer Herholz, K.Katja Heuer, F.Felix Hoffstaedter, R.Ruoqi Huang, R.Renzo Huber, O.Ole Jensen, K.Kan Keeratimahat, J. Q.Julian Q. Kosciessa, S.Sladjana Lukic, N.Neville Magielse, C. J.Christopher J. Markiewicz, C. G.Caroline G. Martin, C.Camille Maumet, A.Anna Menacher, J.Jeff Mentch, C.Christian Mönch, S.Shammi More, L.Leonardo Muller, L.Leonardo Muller-Rodriguez, S. A.Samuel A. Nastase, E.Eliana Nicolaisen-Sobesky, D. M.Dylan M. Nielson, C. R.Christopher R. Nolan, F.François Paugam, P.Pedro Pinheiro-Chagas, A. L.Ana Luísa Pinho, A.Alessandra Pizzuti, B.Benjamin Poldrack, B. A.Benedikt A. Poser, R.Roberta Rocca, J.Jacob Sanz-Robinson, K.Kelvin Sarink, K. R.Kevin R Sitek, N.Nadine Spychala, R.Rüdiger Stirnberg, M.Michał Szczepanik, M.Mohammad Torabi, R.Roberto Toro, S. G.Sebastian G. W. Urchs, S. L.Sofie L. Valk, A. S.Adina S Wagner, L. K.Laura K. Waite, A. Q.Alexander Q. Waite, L.Lea Waller, T. J.Tyler J. Wishard, J.Jianxiao Wu, Y.Yuchen Zhou and J. D.Janine D. Bijsterbosch. Proceedings of the OHBM Brainhack 2022.Aperture NeuroMarch 2024HALDOI

International peer-reviewed conferences

• 24 inproceedingsA. N.Abderaouf Nassim Amalou, E.Elisa Fromont and I.Isabelle Puaut. Fast and Accurate Context-Aware Basic Block Timing Prediction using Transformers.Proceedings of the ACM SIGPLAN 2024 International Conference on Compiler ConstructionCC 2024 - ACM SIGPLAN 33rd International Conference on Compiler ConstructionEdimbourg, United KingdomACM; ACMMarch 2024, 227–237HALDOIback to text
• 25 inproceedingsZ.Zubaria Asma, D.Daniel Hernández, L.Luis Galárraga, G.Giorgos Flouris, I.Irini Fundulaki and K.Katja Hose. NPCS: Native Provenance Computation for SPARQL.WWW 2024 - ACM Web ConferenceSingapore, SingaporeACMMay 2024, 2085 - 2093HALDOIback to text
• 26 inproceedingsS.Sébastien Ferré. Tackling the Abstraction and Reasoning Corpus (ARC) with Object-centric Models and the MDL Principle.LNAI, SpringerIDA 2024 - Symposium on Intelligent Data AnalysisStockholm, Sweden2024, 1-12HALback to text
• 27 inproceedingsE.Elodie Germani, E.Elisa Fromont and C.Camille Maumet. Uncovering communities of pipelines in the task-fMRI analytical space.IEEE International Conference on Image ProcessingAbu Dhabi, United Arab EmiratesOctober 2024HALDOIback to text
• 28 inproceedingsY.Yasmine Hachani, P.Patrick Bouthemy, E.Elisa Fromont, S.Sylvie Ruffini, L.Ludivine Laffont and A.Alline de Paula Reis. Early prediction of the transferability of bovine embryos from videomicroscopy.ICIP 2024 - IEEE International Conference on Image ProcessingAbu DHABI, United Arab EmiratesIEEE2024, 1-6HALback to text
• 29 inproceedingsG.Gwladys Kelodjou, L.Laurence Rozé, V.Véronique Masson, L.Luis Galárraga, R.Romaric Gaudel, M.Maurice Tchuente and A.Alexandre Termier. Shaping Up SHAP: Enhancing Stability through Layer-Wise Neighbor Selection.AAAI 2024 - 38th Annual AAAI Conference on Artificial IntelligenceVancouver, Canada2024, 1-10HALback to text
• 30 inproceedingsT.Thomas Lebrun, L.Louis Béziaud, T.Tristan Allard, A.Antoine Boutet, S.Sébastien Gambs and M.Mohamed Maouche. Synthetic Data: Generate Avatar Data on Demand.The International Web Information Systems Engineering conference (WISE)15440Lecture Notes in Computer ScienceDoha-Qatar, FranceSpringer Nature SingaporeNovember 2025, 193-203HALDOIback to text
• 31 inproceedingsA. É.Alexandra Épiphanie Padonou, P.Peggy Cellier and S.Sébastien Ferré. Investigating the use of language models for natural language querying over knowledge graphs.RNTIEGC 2024 - 24ème conférence francophone sur l'Extraction et la Gestion des ConnaissancesRNTI-E-40Dijon, France2024, 1-8HALback to text
• 32 inproceedingsJ.Jordan Patracone, P.Paul Viallard, E.Emilie Morvant, G.Gilles Gasso, A.Amaury Habrard and S.Stéphane Canu. A Theoretically Grounded Extension of Universal Attacks from the Attacker's Viewpoint.ECML PKDD 2024 - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in DatabasesVilnius, Lithuania2024, 1-27HALDOIback to text
• 33 inproceedingsM. R.Mohammed Rameez Rameez Qureshi, N.Naïm Es-Sebbani, L.Luis Galárraga, Y.Yvette Graham, M.Miguel Couceiro and Z.Zied Bouraoui. REFINE-LM: Mitigating Language Model Stereotypes via Reinforcement Learning.Frontiers in Artificial Intelligence and ApplicationsECAI 2024 - 27th European Conference on Artificial Intelligence392Santiago de Compostela, SpainOctober 2024, 4027-4034HALDOIback to text
• 34 inproceedingsP.Paul Viallard, R.Rémi Emonet, A.Amaury Habrard, E.Emilie Morvant and V.Valentina Zantedeschi. Leveraging PAC-Bayes Theory and Gibbs Distributions for Generalization Bounds with Complexity Measures.Proceedings of The 27th International Conference on Artificial Intelligence and StatisticsAISTATS 2024 - 27th International Conference on Artificial Intelligence and StatisticsValencia, Spain2024HALback to text

National peer-reviewed Conferences

• 35 inproceedingsV.Vanessa Fokou, P.Peggy Cellier, X.Xavier Dolques, S.Sébastien Ferré and F.Florence Le Ber. Comparing Relational Concept Analysis and Graph-FCA on their Common Ground.Concepts 2024, Cadiz, Spain, septembre 2024Cadiz, SpainSeptember 2024HALback to text
• 36 inproceedingsY.Yasmine Hachani, P.Patrick Bouthemy, S.Sylvie Ruffini, L.Ludivine Laffont, E.Elisa Fromont and A. d.Alline de Paula Reis. Prédiction précoce de la transférabilité d'embryons bovins par vidéomicroscopie.RFIAP 2024RFIAP 2024 - Congrès Reconnaissance des Formes, Image, Apprentissage et PerceptionLille, France2024HALback to text

Conferences without proceedings

• 37 inproceedingsH.Hind Atbir, F.Farah Cherfaoui, G.Guillaume Metzler, E.Emilie Morvant and P.Paul Viallard. Une borne PAC-Bayésienne sur une mesure de risque pour l'apprentissage équitable.CAP 2024 - Conférence sur l'Apprentissage AutomatiqueLille, France2024HALback to text
• 38 inproceedingsG. C.G. Charbel N. Kindji, L. M.Lina Maria Rojas-Barahona, E.Elisa Fromont and T.Tanguy Urvoy. Cross-table Synthetic Tabular Data Detection.COLING 2025 Workshop on Detecting AI Generated ContentAbu dahbi, United Arab EmiratesJanuary 2025HAL

Scientific book chapters

• 39 inbookA.Agnieszka Ławrynowicz, L.Luis Galárraga, M.Mehwish Alam, B.Bérénice Jaulmes, V.Václav Zeman and T.Tomas Kliegr. Neurosymbolic Methods for Rule Mining.Handbook on Neurosymbolic Artificial Intelligence2024. In press. HALback to text

Edition (books, proceedings, special issue of a journal)

• 40 proceedingsConceptual Knowledge Structures.CONCEPTS 2024 - 1st International Joint Conference on Conceptual Knowledge Structures14914Lecture Notes in Computer ScienceSpringer Nature Switzerland2024, 1-348HALDOIback to text
• 41 proceedingsSSDBM '24: Proceedings of the 36th International Conference on Scientific and Statistical Database Management.SSDBM 2024 - 36th International Conference on Scientific and Statistical Database ManagementACMAugust 2024, 1-165HALDOIback to textback to textback to text

Doctoral dissertations and habilitation theses

• 42 thesisO.Olivier Gauriau. Numerical rule mining for the prediction of plant disease dynamics..Université de RennesNovember 2024HAL

Reports & preprints

• 43 miscB.Benjamin Dupuis, P.Paul Viallard, G.George Deligiannidis and U.Umut Şimşekli. Uniform Generalization Bounds on Data-Dependent Hypothesis Sets via PAC-Bayesian Theory on Random Sets.April 2024HALback to text
• 44 miscE.Elodie Germani, N.Nikhil Baghwat, M.Mathieu Dugré, R.Rémi Gau, A.Albert Montillo, K.Kevin Nguyen, A.Andrzej Sokolowski, M.Madeleine Sharp, J.-B.Jean-Baptiste Poline and T.Tristan Glatard. Predicting Parkinson's disease trajectory using clinical and functional MRI features: a reproduction and replication study.September 2024HALback to text
• 45 miscE.Elodie Germani, E.Elisa Fromont, P.Pierre Maurel and C.Camille Maumet. The HCP multi-pipeline dataset: an opportunity to investigate analytical variability in fMRI data analysis.2024HALback to text
• 46 miscE.Elodie Germani, C.Camille Maumet and E.Elisa Fromont. Mitigating analytical variability in fMRI results with style transfer.2024HALback to text
• 47 miscM.Maxime Haddouche, P.Paul Viallard, U.Umut Şimşekli and B.Benjamin Guedj. A PAC-Bayesian Link Between Generalisation and Flat Minima.February 2024HALback to text
• 48 miscG. C.G. Charbel N. Kindji, L. M.Lina Maria Rojas-Barahona, E.Elisa Fromont and T.Tanguy Urvoy. Under the Hood of Tabular Data Generation Models: Benchmarks with Extensive Tuning.2024HAL
• 49 miscP.Paul Viallard, M.Maxime Haddouche, U.Umut Şimşekli and B.Benjamin Guedj. Tighter Generalisation Bounds via Interpolation.2024HALDOIback to text

Other scientific publications

• 50 inproceedingsY.Yasmine Hachani, P.Patrick Bouthemy, S.Sylvie Ruffini, L.Ludivine Laffont, E.Elisa Fromont and A.Alline de Paula Reis. Early prediction of the transferability of bovine embryos from videomicroscopy.IABM 2024 - Colloque Français d'Intelligence Artificielle en Imagerie BiomédicaleGrenoble, France2024, 1-1HALback to textback to text

Scientific popularization

• 51 miscF.Fanny Dégeilh, E.Elodie Germani, N.Nolwenn Jégou, C.Carla Joud, J.Jeremy Lefort-Besnard and C. O.Camille O Muller. À la découverte des métiers de la recherche en imagerie cérébrale.Rennes, FranceMarch 2024, 1-89HAL