Portrait de Pablo Piantanida

Pablo Piantanida

Membre académique associé
Professeur titulaire, Université Paris-Saclay
Directeur, Laboratoire international de recherche sur les systèmes d'apprentissage (ILLS) du CNRS, Director of the CNRS International Research Laboratory on Learning Systems (ILLS)
Professeur associé, McGill University University and École de technologie supérieure (ETS), Département de génie des systèmes
Sujets de recherche
Sécurité de l'IA
Systèmes multi-agents
Théorie de l'apprentissage automatique
Théorie de l'information
Traitement du langage naturel

Biographie

Pablo Piantanida est professeur titulaire à CentraleSupélec | Université Paris-Saclay et directeur du Laboratoire international de recherche sur les systèmes d'apprentissage (ILLS) du CNRS, situé à l'ÉTS Montréal. Il est également professeur associé à l'Université McGill et à l'ÉTS Montréal.

Axe de recherche : Fondements mathématiques et de la théorie de l'information d'une intelligence fiable, adaptative et collaborative.

Domaines de recherche : Théorie de l'information, théorie de l'apprentissage automatique, sûreté et fiabilité de l'IA, théorie statistique de la décision, incertitude et évaluation, apprentissage de représentations, systèmes multi-agents, protection de la vie privée et traitement automatique du langage naturel (TALN).

Comment un système d'IA devrait-il décider quelle information apprendre ou acquérir, à quelles données probantes se fier, à quel moment communiquer et quand s'abstenir ? Pablo Piantanida explore ces questions au croisement de la théorie de l'information, de la statistique, de la théorie de la décision et de l'apprentissage automatique.

Ses travaux portent notamment sur la caractérisation des limites de communication dans l'inférence distribuée, où des agents prennent des décisions à partir d'informations partielles ; l'élaboration d'approches permettant de comparer les représentations en fonction des décisions qu'elles soutiennent ; ainsi que la conception de méthodes pour détecter les erreurs de prédiction et quantifier l'incertitude en classification et en régression. Il étudie également la manière d'auditer les systèmes d'IA avec des budgets de test restreints et le recours aux approches issues de la théorie de l'information pour les évaluer.

Son programme de recherche s'articule autour de trois grands axes : l'apprentissage, la représentation et le raisonnement ; l'incertitude, l'évaluation et la fiabilité ; ainsi que l'intelligence collaborative et agentique. À travers ces avenues, il cherche à déterminer à partir de quel seuil un ajout de données, de calcul ou d'interaction améliore suffisamment une décision pour en justifier le coût. Ces recherches offrent aux étudiantes et étudiants ainsi qu'aux partenaires industriels l'occasion de relier la théorie fondamentale au déploiement de systèmes d'IA fiables en contexte réel.

Étudiants actuels

Doctorat - McGill
Superviseur⋅e principal⋅e :
Doctorat - McGill
Superviseur⋅e principal⋅e :
Doctorat - École de technologie suprérieure
Maîtrise recherche - École de technologie suprérieure
Postdoctorat - École de technologie suprérieure
Co-superviseur⋅e :
Doctorat - Université Paris-Saclay
Doctorat - McGill
Superviseur⋅e principal⋅e :
Doctorat - Université Paris Dauphine-PSL
Maîtrise recherche - École de technologie suprérieure
Co-superviseur⋅e :
Collaborateur·rice de recherche - Sorbonne Université

Publications

ECUAS: A family of metrics for principled evaluation of uncertainty-augmented systems
Lautaro Estienne
Erik Ernst
Matias Vera
LUCIANA FERRER
In high-stakes automated decision-making, access to predictive uncertainty is essential for enabling users -- human or downstream systems --… (voir plus) to accept or reject predictions based on application-specific cost trade-offs. Such uncertainty-augmented (UA) systems -- i.e., systems that output both predictions and uncertainty scores -- are currently being assessed in the literature in a variety of ways, using separate metrics to evaluate the predictions and the uncertainty scores, setting a cost function with a fixed rejection cost or integrating over a coverage-risk curve. We argue that these evaluation approaches are inadequate for assessing overall performance of the UA system for decision making under uncertainty and propose a novel family of metrics, ECUAS
QueStER: Query Specification for Generative Keyword-Based Retrieval
Yuxuan Zong
Habiboulaye Amadou Boubacar
Benjamin Piwowarski
Generative retrieval (GR) differs from the traditional index–then–retrieve pipeline by storing relevance in model parameters and generat… (voir plus)ing retrieval cues directly from the query, but it can be brittle out of domain and expensive to scale. We introduce QueStER (QUEry SpecificaTion for gEnerative Keyword-Based Retrieval), which bridges GR and query reformulation by learning to generate explicit keyword-based search specifications. Given a user query, a lightweight LLM produces a keyword query that is executed by a standard retriever (BM25), combining the generalization benefits of generative query rewriting with the efficiency and scalability of lexical indexing. We train the rewriting policy with reinforcement learning techniques. Across in- and out-of-domain evaluations, QueStER consistently improves over BM25 and is competitive with neural IR baselines, while maintaining strong efficiency.
Auditing is not Evaluating: LLM Audit Requires Dynamic, Contextual, Budget-aware and Reliable Evidence
DynaPFN: Zero-Shot Dynamical System Forecasting with Tabular Prior-Fitted Networks
From Infrastructure to Interface, the AI Value Chain Drives LLM Homogenization
Source-Free Class Relearning: Diagnosing Forgetting in Class Unlearning
Zahra Dehghani
Mohammadhadi Shateri
Class unlearning aims to remove a model's ability to recognize designated forget classes while preserving performance on retain classes. How… (voir plus)ever, low forget accuracy after unlearning does not necessarily mean the class structure has been erased. Approximate unlearning methods can alter classifier decision boundaries while leaving recoverable structure in the representation. Prior work has shown that forget classes can be recovered, but existing approaches require real forget or retain samples, auxiliary data, or reference checkpoints. We study class relearning in a strictly source-free setting, asking whether a forget class can be recovered through a classifier-head update using only the unlearned model. Our approach rests on a theoretical analysis establishing a sufficient alignment condition under which a single gradient step on a synthetic probe set increases the expected logit margin of the forget class. Building on this, we propose a white-box Source-Free Relearning Audit (SFRA), which generates candidate embeddings in representation space and uses model-guided confidence filtering to construct high-confidence retain probes and low-confidence boundary-adjacent probes that are relabelled as the forget class. Gaussian sampling and Softmax confidence are used by default, while ablations with alternative proposal distributions and uncertainty criteria show that recoverability is not specific to these choices. To quantify recoverability, we introduce the Relearning Score (RS), which jointly measures forget-class recovery and retain-accuracy preservation, and report class-matched
A Universal Source-Free Class Unlearning Framework via Synthetic Embeddings
Mohammadhadi Shateri
Class unlearning in neural classifiers refers to selectively removing the model’s ability to recognize a target (forget) class by reshapin… (voir plus)g the decision boundaries. This is essential when taxonomies change, labels are corrected, or legal or ethical requirements mandate class removal. The objective is to preserve performance on the remaining (retain) classes while avoiding costly full retraining. Existing methods generally require access to the source, i.e., forget/retain data or a relevant surrogate dataset. This dependency limits their applicability in scenarios where access to source data is restricted or unavailable. Even the recent source-free class unlearning methods rely on generating samples in the data space, which is computationally expensive and not even essential for doing class unlearning. In this work, we propose a novel source-free class unlearning framework that enables existing unlearning methods to operate using only the deployed model. We show that, under assumptions on the forget loss with respect to logits, class unlearning can be performed source-free for any given neural classifier by utilizing randomly generated samples within the classifier’s intermediate space. Specifically, randomly generated embeddings pseudo-labeled by the model as belonging to the forget or retain classes can support effective source-free unlearning. Our analysis further shows that, under conditions on the forget loss and synthetic forget embeddings, minimizing the forget loss induces expected logit shifts consistent with class unlearning, without requiring a specific parametric form of the embedding distribution. We validate our framework on four backbone architectures, ResNet-18, ResNet-50, ViT-B/16, and Swin-T, across three benchmark datasets, CIFAR-10, CIFAR-100, and TinyImageNet. Our experimental results show that existing class unlearning methods can operate within our source-free framework, with minimal impact on their forgetting efficacy and retain class accuracy. The code is available at https://github.com/Yasaman-dt/Source_Free_Class_Unlearning.
MolRGen: A Training and Evaluation Setting for De Novo Molecular Generation with Reasonning Models
Recent advances in reasoning-based large language models (LLMs) have demonstrated substantial improvements in complex problem-solving tasks.… (voir plus) Motivated by these advances, several works have explored the application of reasoning LLMs to drug discovery and molecular design. However, most existing approaches either focus on evaluation or rely on training setups that require ground-truth labels, such as molecule pairs with known property modifications. Such supervision is unavailable in \textit{de novo} molecular generation, where the objective is to generate novel molecules that optimize a desirability score without prior knowledge of high-scoring candidates. To bridge this gap, we introduce MolRGen, a large-scale benchmark and dataset for training and evaluating reasoning-based LLMs on \textit{de novo} molecular generation. Our contributions are threefold. First, we propose a setting to evaluate and train models for \textit{de novo} molecular generation and property prediction. Second, we introduce a novel diversity-aware top-
Position: Auditing Is Not Evaluating; LLM Audit Requires Dynamic, Contextual, Budget-Aware and Reliable Evidence
Auditing large language models (LLMs) is increasingly urgent as these systems are deployed in high-stakes settings, yet existing evaluation … (voir plus)practices are ill-suited to meet auditing requirements. Directly repurposing standard evaluation tools can yield incomplete or misleading conclusions, e.g. overstating robustness when evidence comes from static prompts rather than adaptive, real-world interactions. This position paper argues that effective LLM audits must instead generate dynamic, context-sensitive, budget-aware, and reliable evidence. To support this position, we analyze how each of these principles can be operationalized through a four-component framework: Auditing Scope, Interactor, Evaluator, and Output. We highlight design requirements, assumptions, limitations and research directions, demonstrating how high-level principles can be translated into concrete, actionable, evidence-based procedures.
Happiness as a Measure of Fairness
Georg Pichler
Marco Romanelli
In this paper, we propose a novel fairness framework grounded in the concept of _happiness_, a measure of the utility each group gains from … (voir plus)decision outcomes. By capturing fairness through this intuitive lens, we not only offer a more human-centered approach, but also one that is mathematically rigorous: In order to compute the optimal, fair post-processing strategy, only a linear program needs to be solved. This makes our method both efficient and scalable with existing optimization tools. Furthermore, it unifies and extends several well-known fairness definitions, and our empirical results highlight its practical strengths across diverse scenarios.
An Indicator of Membership Inference Security in Post-Training Quantized Models
Quantizing machine learning models has demonstrated its effectiveness in lowering memory and inference costs while maintaining performance l… (voir plus)evels comparable to those of the original models. In this work, we investigate the impact of quantization procedures on privacy in data-driven models, focusing on their vulnerability to membership inference attacks. Membership Inference Security (MIS) has recently been proposed to characterize the privacy of machine learning models against the most powerful (and possibly unknown) attacks. However, quantifying MIS appears to be computationally very difficult. In this paper, we propose a new MIS indicator for post-training quantization procedures of machine learning models that minimize an empirical loss. This new indicator is a byproduct of a theoretical asymptotic analysis of the MIS in this context. We also present a methodology for empirically estimating our MIS indicator. Using synthetic datasets and real-world data (in the context of drug discovery), we demonstrate the effectiveness of our approach in assessing and ranking the MIS of different quantizers.
Adapting Language Models to Produce Good Class Probabilities for Classification Tasks
Lautaro Estienne
Matias Vera
Elizabeth Fons
Elena Kochkina
LUCIANA FERRER
Large generative language models (GLM) provide a versatile tool for solving a wide variety of natural processing tasks. GLM responses, thoug… (voir plus)h, are provided in the form of text, without an indication of the model's confidence in the answer. This limits the usability of these models on high-risk applications where decisions made based on an incorrect answer can have severe consequences. In this work, we focus on the problem of generating class posterior distributions for text classification tasks like sentiment, news category and intent classification. These posteriors can be used for decision making and as interpretable scores for the user. We show that the naive approach for computing posteriors based on the token posteriors produced by the GLM results in extremely poor posteriors. We then explore different adaptation approaches for improving the quality of posteriors, focusing on low resource scenarios where a small amount of data is available for adaptation. We show that parameter-efficient supervised fine-tuning (SFT), while providing large gains in terms of decision quality, produces suboptimal posteriors due to overfitting. To address this problem, we propose an approach that combines SFT and post-hoc calibration (PHC) using a three-stage training strategy, improving the quality of both posteriors and categorical decisions.