|
|
|
Program > Program CAp
Monday
Invited Speaker 1 (Juliette Sénéchal) 9:15 - 10:15
AI Alignment Through EU AI Act
The launch of Chat GPT, a general-purpose generative AI system, on the global market in November 2022 has led to a proliferation of reflections on the new risks posed to humans by artificial intelligence systems. In particular, these reflections have highlighted the fact that generative AI, as a substitute for human language and human creativity, runs the risk of exposing humans to forms of cultural uniformisation and disinformation by deepfakes, undermining the ability of humans to compose a society and to adapt themselves to the constraints of their territory. In the context of these reflections, it is possible to observe a proliferation of soft and hard norms emanating from companies, governments on different continents, supranational organisations, researchers, etc., aimed at bringing AI systems into alignment with human values. This proliferation of norms raises three questions: what human values should be protected at a time when generative AI is taking a technological leap forward? What types of norms should be prioritised to implement this protection effectively: ethical norms, legal acts, technical standardisation? What territorial scope should these norms have? This contribution will study the normative approaches proposed by the future European regulation on artificial intelligence (the AI Act) in order to address these questions.
Session 1 (Fairness ~ Interpretability) 10:45 - 12:00
Chair: Romaric Gaudel
| Title | Authors | Presentation length |
|---|---|---|
| Assessing Uncertainty in Similarity Scoring: Performance & Fairness in Face Recognition | Jean-Rémy Conti, Stephan Clémençon | 15 min |
| Impact de la Perturbation des Prédictions sur l’Équité en Classification Binaire et Linéaire | Vitalii Emelianov, Michaël Perrot | 5 min |
| Une borne PAC-Bayésienne sur une mesure de risque pour l'apprentissage équitable | Hind Atbir, Farah Cherfaoui, Guillaume Metzler, Emilie Morvant, Paul Viallard | 5 min |
| Apprentissage de Modèles Équitables via Repondération en Apprentissage Fédéré | Paul Andrey, Brahim Erraji, Michaël Perrot | 5 min |
| Augment to Interpret: Unsupervised and Inherently Interpretable Graph Embeddings | Gregory Scafarto, Madalina Ciortan, Simon Tihon, Quentin Ferré | 15 min |
| Affiner SHAP : Améliorer la stabilité grâce à la sélection de voisins en couches | Gwladys Kelodjou, Laurence Rozé, Véronique Masson, Luis Galárraga, Romaric Gaudel, Maurice Tchuente, Alexandre Termier | 5 min |
| Weighted majority vote using Shapley values in crowdsourcing | T. Lefort, B. Charlier, A. Joly, J. Salmon | 5 min |
| Exploration de la sémantique dans l’attention d’un modèle de langue pré-entraîné | Frédéric Charpentier, Adrien Guille, Jairo Cugliari | 5 min |
Invited Speaker 2 (Celestine Mendler-Dünner) 13:30 - 14:30
Performativity in Machine Learning
Algorithmic predictions create expectations, inform decisions and steer consumption. As such, predictions used in societal systems not only describe the population they aim to predict, but they have the power to change it; a prevalent phenomenon often neglected in theories and practices of machine learning. In this talk, I will start by introducing the calculus of performative prediction, that conceptualizes this phenomenon by allowing the predictive model to influence the distribution over future data. This dynamic perspective on prediction elucidates new solution concepts, optimization challenges, and brings forth interesting connections to concepts from causality and game theory that I will discuss. Moreover, it allows us to articulate two mechanisms fundamental to prediction, learning and steering, making explicit the important role of power in prediction, and giving us a means to measure it. I will end my talk on a discussion of collective action as a strategy to resist the power of digital platforms.
Session 2 (Optimal Transport ~ Learning Theory) 14:30 - 16:00
Chair: Gilles Gasso
| Title | Authors | Presentation length |
|---|---|---|
| A framework for deep Supervised Graph Prediction | Paul Krzakala, Junjie Yang, Rémi Flamary, Florence D'Alché, Charlotte Laclau, Matthieu Labeau | 15 min |
| Learning via Wasserstein-Based High Probability Generalisation Bounds | Paul Viallard, Maxime Haddouche, Umut Simsekli, Benjamin Guedj | 15 min |
| Exploiting Edge Features in Graph-based Learning with Fused Network Gromov-Wasserstein Distance | Junjie Yang, Matthieu Labeau, Florence d'Alché-Buc | 5 min |
| Détection non supervisée d'anomalies dans les images satellites pour le monitoring des surfaces océaniques par une composition d'ACP robuste et d'un test d'adequation sur la distance de Wasserstein entre processus ponctuels | Julien Bastian, Stephane Chretien, Ben Gao, Rémi Vaucher | 5 min |
| Régularisation implicite dans la décomposition de Tucker sur-paramétrée : parcimonie structurée et approximation de rang faible | Kais HARIZ, Hachem Kadri, Stéphane Ayache, Maher Moakher, Thierry Artières | 15 min |
| Bornes en Généralisation PAC-Bayes pour RNN Simples ne dépendant pas de la longueure des Séquences | Volodimir Mitarchuk, Clara Lacroce, Eyraud Rémi, Rémi Emonet, Amaury Habrard, Guillaume Rabusseau | 5 min |
| A Symmetry-Aware Exploration of Bayesian Neural Network Posteriors | Olivier Laurent, Emanuel Aldea, Gianni Franchi | 5 min |
| Active Fourier Auditor for Estimating Distributional Properties of Machine Learning Models | Ayoub Ajarra, Bishwamittra Ghosh, Debabrota Basu | 5 min |
Session 3 (Decentralized Learning ~ Privacy ~ Adversarial Attacks) 16:30 - 17:45
Chair: Marc Tommasi
| Title | Authors | Presentation length |
|---|---|---|
| Apprentissage Privé Décentralisé par Marche Aléatoire | Edwige Cyffers, Aurélien Bellet, Jalaj Upadhyay | 15 min |
| Reconstruction de données en Apprentissage Décentralisé | Abdellah El Mrini, Edwige Cyffers, Aurélien Bellet | 5 min |
| Apprentissage Fédéré pour Inférences Causales : Estimer l'Effet Traitement avec des Données Décentralisées | Rémi Khellaf, Aurélien Bellet, Julie Josse | 5 min |
| Apprentissage décentralisé respectueux de la vie privée à faible coût | Sayan Biswas, Davide Frey, Romaric Gaudel, Anne-Marie Kermarrec, Dimitri Lerévérend, Rafael Pires, Rishi Sharma, François Taïani | 5 min |
| On the Complexity of Differentially Private Best-Arm Identification with Fixed Confidence | Achraf Azize, Marc Jourdan, Aymen Al Marjani, Debabrota Basu | 15 min |
| DP-SGD with weight clipping | Antoine Barczewski, Jan Ramon | 5 min |
| Neural ODE à divergence nulle pour la classification | Zakaria Jarraya, Simon Benaïchouche, Lucas Drumetz, Douraied Ben Salem, François Rousseau | 5 min |
| Confidentialité Pufferfish de Rényi : Mécanismes Additifs Généraux et Amplification de Confidentialité par Itération | Clément Pierquin, Aurélien Bellet, Marc Tommasi, Matthieu Boussard | 5 min |
Discussion around Reviews and LLMs 17:45 - 18:15
Session Poster 1 18:15 - 19:30
Tuesday
Prix de Thèse SSFAM/AFRIF 9:00 - 10:15
| Title | Authors | Presentation length |
|---|---|---|
| Remise des prix | SSFAM, AFRIF | 15 min |
| Towards Securing Machine Learning Algorithms through Misclassification Detection and Adversarial Attack Detection | Federica Granese | 5 min |
| Combinaison de réseaux de régulation génique et d’apprentissage statistique séquentiel pour le repositionnement de médicaments | Clémence Réda | 5 min |
| Diverse and Efficient Ensembling of Deep Networks | Alexandre Ramé | 15 min |
| Efficient Neural Networks: Post Training Pruning and Quantization | Édouard Yvinec | 25 min |
Joint Session with RFIAP 1 10:45 - 12:00
| Title | Authors | Presentation length |
|---|---|---|
| Random matrix analysis to balance between supervised and unsupervised learning under the low density separation assumption | Vasilii Feofanov, Malik Tiomoko, Aladin Virmaux | 15 min |
| Find the Lady: Permutation and Re-Synchronization of Deep Neural Networks | Carl De Sousa Trias, Mihai Mitrea, Attilio Fiandrotti, Marco Cagnazzo, sumanta chaudhuri, Enzo Tartaglione | 15 min |
| Inverse problem regularization with hierarchical variational autoencoders | Jean Prost, Antoine Houdard, Andres Almansa, Nicolas Papadakis | 15 min |
| Sequential Representation Learning via static-dynamic Conditional Disentanglement | Mathieu Cyrille Simon, Pascal Frossard, Christophe De Vleeschouwer | 15 min |
Invited Speaker 3 (Matthieu Cord) 13:30 - 14:30
Vision and Language with transformers
Since the seminal publication on transformers in Natural Langage Processing, the Computer Vision community has rapidly adopted and adapted this model to a multitude of vision tasks. In this seminar, I will trace these major developments, starting with image classification. I will then explore various extensions, such as object detection, image segmentation and captioning. In particular, this will enable us to illustrate how transformers have revolutionized the modeling of Vision and Language relationships.
Session 4 (Graphs ~ Optimization) 14:30 - 16:00
Chair: Charlotte Laclau
| Title | Authors | Presentation length |
|---|---|---|
| Modélisation continue des séries temporelles pour l'imputation et la prévision avec des représentations neuronales implicites | Etienne Le Naour, Louis Serrano, Léon Migus, Yuan Yin, Ghislain Agoua, Nicolas Baskiotis, Patrick Gallinari, Vincent Guigue | 15 min |
| SANGEA: Scalable and Attributed Network Generation | Valentin Lemaire, Youssef Achenchabe, Lucas Ody, Houssem Eddine Souid, Gianmarco Aversano, Nicolas Posocco, Sabri Skhiri | 5 min |
| Sur l'amélioration des transformers pour la prévision de séries temporelles en évitant les mauvais minima locaux | Romain Ilbert, Ambroise Odonnat, Vasilii Feofanov, Aladin Virmaux, Giuseppe Paolo, Themis Palpanas, Ievgen Redko | 5 min |
| Graph-Survival: A Survival Analysis Framework for Machine Learning on Temporal Networks | Mathilde Perez, Raphael Romero, Bo Kang, Tijl De Bie, Jefrey Lijffijt, Charlotte Laclau | 5 min |
| Sketch In, Sketch Out: Accelerating both Learning and Inference for Structured Prediction with Kernels | Tamim El Ahmad, Luc Brogat-Motte, Pierre Laforgue, Florence d'Alché-Buc | 15 min |
| Leveraging Ensemble Diversity for Robust Self-Training in the Presence of Sample Selection Bias | Ambroise Odonnat, Vasilii Feofanov, Ievgen Redko | 15 min |
| Support Exploration Algorithm: estimateurs directs et reconstruction parcimonieuse | Mimoun Mohamed, Francois Malgouyres, Valentin Emiya, Caroline Chaux | 5 min |
| SCAFFLSA: Quantification et élimination du biais dû à l'hétérogénéité en approximation stochastique linéaire et TD learning | Paul Mangold, Sergey Samsonov, Safwan Labbi, Ilya Levin, Reda ALAMI, Alexey Naumov, Eric Moulines | 5 min |
| Preserving Calibration by Tailoring Mixup to Data | Quentin Bouniot, Pavlo Mozharovskyi, Florence d'Alché-Buc | 5 min |
Session 5 (Machine Learning and its Applications) 16:30 - 17:45
Chair: Vincent Guigue
| Title | Authors | Presentation length |
|---|---|---|
| Apprentissage automatique informé par la physique : vers une meilleure compréhension de l’interaction laser-matière | Fayad Ali Banna, Jean-Philippe Colombier, Rémi Emonet, Marc Sebban | 15 min |
| Erreur d’Approximation pour les Fonctions Sobolev Regulières avec des Réseaux de Neurones tanh : Impact Theorique sur les PINNs | Benjamin Girault, Rémi Emonet, Amaury Habrard, Jordan Patracone, Marc Sebban | 5 min |
| Bregman Fourier Neural Operators | Abdel-Rahim MEZIDI, Rémi Emonet, Amaury Habrard, Jordan PATRACONE, massimiliano pontil, Saverio Salzo, Marc Sebban | 5 min |
| Domain Translation via Latent Space Mapping | Tsiry Mayet, Simon Bernard, Clement Chatelain, Romain Herault | 15 min |
| RelevAI-Reviewer: A Benchmark on AI Reviewers for Survey Paper Relevance | Paulo Henrique Couto, Quang Phuoc Ho, Nageeta Kumari, Benedictus Kent Rachmat, Thanh Gia Hieu Khuong, Ihsan Ullah, Lisheng Sun-Hosoya | 5 min |
| When Quantization Affects Confidence of Large Language Models? | Irina Proskurina, Luc Brun, Guillaume Metzler, Julien Velcin | 5 min |
| Fine-grained probing of foundation models in the auditory modality | Etienne Bost, Mitsuko Aramaki, Richard Kronland-Martinet, Sølvi Ystad, Thierry Artières, Thomas Schatz | 5 min |
Assemblée Générale 17:45 - 18:30
Session Poster 2 18:30 - 19:30
Wednesday
Invited Speaker 4 (Xavier Alameda-Pineda) 9:00 - 10:00
Learning for Companion Robots: Preparation and Adaptation
Companion and social robots are a chimeric mirage. On the one side, they are part of our collective representation of ready-to-deploy technological developments. On the other side, there are a series of scientific and technological barriers to such deployment. One of the scientific challenges is the difficulty of such robots to successfully perform in the wide variety of environments and social situations they need to face. In this talk, a series of works developed at the RobotLearn team at Inria at Univ. Grenoble Alpes will be presented, all aligned with the philosophy of developing learning strategies that would allow a social/companion robotic platform to quickly adapt its perception and action models to unseen situations. Applications such as multi-person tracking, audio-visual speech modeling, and navigation policies will be discussed.
Joint Session with RFIAP 2 10:30 - 12:00
Chair: Elisa Fromont
| Title | Authors | Presentation length |
|---|---|---|
| NECO: NEural Collapse Based Out-of-distribution detection | Mouïn Ben Ammar, Nacim Belkhir, Sebastian Popescu, Antoine Manzanera, Gianni Franchi | 15 min |
| Mining bias-target Alignment from Voronoi Cells | Remi Nahon, Van-Tam Nguyen, Enzo Tartaglione |
15 min |
|
Perceptual Scales Predicted by Fisher Information Metrics |
Jonathan Vacher, Pascal Mamassian |
15 min |
|
Collaborating Foundation Models for Domain Generalized Semantic Segmentation |
Yasser Benigmim, Subhankar Roy, Slim Essid, Vicky Kalogeiton, Stéphane Lathuilière |
15 min |
| Online user: 2 | Privacy |
|
