no code implementations • 12 Dec 2024 • Pablo Morales-Álvarez, Stergios Christodoulidis, Maria Vakalopoulou, Pablo Piantanida, Jose Dolz
However, CLIP adapters have been developed to target discriminative performance, and the quality of their uncertainty estimates has been overlooked.
1 code implementation • 11 Sep 2024 • Matthieu Dubois, François Yvon, Pablo Piantanida
The dissemination of Large Language Models (LLMs), trained at scale, and endowed with powerful text-generating abilities has vastly increased the threats posed by generative AI technologies by reducing the cost of producing harmful, toxic, faked or forged content.
1 code implementation • 23 Jun 2024 • Eduardo Dadalto, Florence Alberge, Pierre Duhamel, Pablo Piantanida
Notably, our framework is easily extensible for future developments in detection scores and stands as the first to combine decision boundaries in this context.
Out-of-Distribution Detection
Out of Distribution (OOD) Detection
no code implementations • 20 Jun 2024 • Florence Regol, Joud Chataoui, Bertrand Charpentier, Mark Coates, Pablo Piantanida, Stephan Gunnemann
Machine learning models can solve complex tasks but often require significant computational resources during inference.
no code implementations • 11 Jun 2024 • Andres Altieri, Marco Romanelli, Georg Pichler, Florence Alberge, Pablo Piantanida
This paper tackles the challenge of detecting unreliable behavior in regression algorithms, which may arise from intrinsic variability (e. g., aleatoric uncertainty) or modeling errors (e. g., model uncertainty).
no code implementations • 11 Jun 2024 • Maxime Darrin, Philippe Formont, Ismail Ben Ayed, Jackie CK Cheung, Pablo Piantanida
Embedders play a central role in machine learning, projecting any object into numerical representations that can, in turn, be leveraged to perform various downstream tasks.
1 code implementation • 11 Jun 2024 • Maxime Darrin, Ines Arous, Pablo Piantanida, Jackie CK Cheung
In this paper, we introduce \sys, a summarization method designed to offer a concise yet comprehensive overview of scholarly reviews.
no code implementations • 2 Apr 2024 • Philippe Formont, Hugo Jeannin, Pablo Piantanida, Ismail Ben Ayed
Few-shot learning has recently attracted significant interest in drug discovery, with a recent, fast-growing literature mostly involving convoluted meta-learning strategies.
no code implementations • 29 Feb 2024 • Maxime Darrin, Philippe Formont, Jackie Chi Kit Cheung, Pablo Piantanida
Assessing the quality of summarizers poses significant challenges.
1 code implementation • 24 Feb 2024 • Federica Granese, Marco Romanelli, Pablo Piantanida
We approach this defensive strategy with utmost caution, operating in an environment where the defender possesses significantly less information compared to the attacker.
1 code implementation • 5 Jan 2024 • Roberta Hansen, Matias Vera, Lautaro Estienne, Luciana Ferrer, Pablo Piantanida
This paper deals with the convergence analysis of the SUCPA (Semi Unsupervised Calibration through Prior Adaptation) algorithm, defined from a first-order non-linear difference equations, first developed to correct the scores output by a supervised machine learning classifier.
no code implementations • 6 Nov 2023 • Mohammadhadi Shateri, Francisco Messina, Fabrice Labeau, Pablo Piantanida
In the present work, the overfitting in GANs is studied in terms of the discriminator, and a more general measure of overfitting based on the Bhattacharyya coefficient is defined.
no code implementations • 21 Oct 2023 • Pierre Colombo, Nathan Noiry, Guillaume Staerman, Pablo Piantanida
One of the pursued objectives of deep learning is to provide tools that learn abstract representations of reality from the observation of multiple contextual situations.
no code implementations • 21 Oct 2023 • Pierre Colombo, Victor Pellegrain, Malik Boudiaf, Victor Storchan, Myriam Tami, Ismail Ben Ayed, Celine Hudelot, Pablo Piantanida
First, we introduce a scenario where the embedding of a pre-trained model is served through a gated API with compute-cost and data-privacy constraints.
1 code implementation • 21 Oct 2023 • Pierre Colombo, Marine Picot, Nathan Noiry, Guillaume Staerman, Pablo Piantanida
The landscape of available textual adversarial attacks keeps growing, posing severe threats and raising concerns regarding the deep NLP system's integrity.
no code implementations • 20 Oct 2023 • Eric Aubinais, Elisabeth Gassiat, Pablo Piantanida
Membership inference attacks (MIA) can reveal whether a particular data point was part of the training dataset, potentially exposing sensitive information about individuals.
1 code implementation • 13 Jul 2023 • Lautaro Estienne, Luciana Ferrer, Matías Vera, Pablo Piantanida
These models are usually trained with a very large amount of unsupervised text data and adapted to perform a downstream natural language task using methods like fine-tuning, calibration or in-context learning.
1 code implementation • 6 Jun 2023 • Eduardo Dadalto, Pierre Colombo, Guillaume Staerman, Nathan Noiry, Pablo Piantanida
A key feature of out-of-distribution (OOD) detection is to exploit a trained neural network by extracting statistical patterns and relationships through the multi-layer classifier to detect shifts in the expected input data distribution.
1 code implementation • 2 Jun 2023 • Eduardo Dadalto, Marco Romanelli, Georg Pichler, Pablo Piantanida
Misclassification detection is an important problem in machine learning, as it allows for the identification of instances where the model's predictions are unreliable.
no code implementations • 20 Feb 2023 • Maxime Darrin, Guillaume Staerman, Eduardo Dadalto Câmara Gomes, Jackie CK Cheung, Pablo Piantanida, Pierre Colombo
More importantly, we show that the usual choice (the last layer) is rarely the best one for OOD detection and that far better results could be achieved if the best layer were picked.
no code implementations • 4 Feb 2023 • Federica Granese, Marco Romanelli, Siddharth Garg, Pablo Piantanida
Multi-armed adversarial attacks, in which multiple algorithms and objective loss functions are simultaneously used at evaluation time, have been shown to be highly successful in fooling state-of-the-art adversarial examples detectors while requiring no specific side information about the detection mechanism.
1 code implementation • CVPR 2023 • Malik Boudiaf, Etienne Bennequin, Myriam Tami, Antoine Toubhans, Pablo Piantanida, Céline Hudelot, Ismail Ben Ayed
We tackle the Few-Shot Open-Set Recognition (FSOSR) problem, i. e. classifying instances among a set of classes for which we only have a few labeled samples, while simultaneously detecting instances that do not belong to any known class.
1 code implementation • 19 Dec 2022 • Nuno M. Guerreiro, Pierre Colombo, Pablo Piantanida, André F. T. Martins
Neural machine translation (NMT) has become the de-facto standard in real-world machine translation applications.
no code implementations • 18 Dec 2022 • Maxime Darrin, Pablo Piantanida, Pierre Colombo
In this work, we focus on leveraging soft-probabilities in a black-box framework, i. e. we can access the soft-predictions but not the internal states of the model.
no code implementations • 24 Nov 2022 • Pierre Colombo, Eduardo D. C. Gomes, Guillaume Staerman, Nathan Noiry, Pablo Piantanida
Deep learning methods have boosted the adoption of NLP systems in real-life applications.
no code implementations • 31 Aug 2022 • Pierre Colombo, Maxime Peyrard, Nathan Noiry, Robert West, Pablo Piantanida
Automatic evaluation metrics capable of replacing human judgments are critical to allowing fast development of new methods.
1 code implementation • 30 Jun 2022 • Federica Granese, Marine Picot, Marco Romanelli, Francisco Messina, Pablo Piantanida
Detection of adversarial examples has been a hot topic in the last years due to its importance for safely deploying machine learning algorithms in critical applications.
1 code implementation • 18 Jun 2022 • Malik Boudiaf, Etienne Bennequin, Myriam Tami, Celine Hudelot, Antoine Toubhans, Pablo Piantanida, Ismail Ben Ayed
Through extensive experiments spanning 5 datasets, we show that OSTIM surpasses both inductive and existing transductive methods in detecting open-set instances while competing with the strongest transductive methods in classifying closed-set instances.
no code implementations • ACL 2022 • Pierre Colombo, Guillaume Staerman, Nathan Noiry, Pablo Piantanida
When working with textual data, a natural application of disentangled representations is fair classification where the goal is to make predictions without being biased (or influenced) by sensitive attributes that may be present in the data (e. g., age, gender or race).
1 code implementation • NeurIPS 2021 • Olivier Veilleux, Malik Boudiaf, Pablo Piantanida, Ismail Ben Ayed
Transductive inference is widely used in few-shot learning, as it leverages the statistics of the unlabeled query set of a few-shot task, typically yielding substantially better performances than its inductive counterpart.
1 code implementation • 30 Mar 2022 • Georg Pichler, Marco Romanelli, Leonardo Rey Vega, Pablo Piantanida
Federated Learning is expected to provide strong privacy guarantees, as only gradients or model parameters but no plain text training data is ever exchanged either between the clients or between the clients and the central server.
1 code implementation • CVPR 2022 • Ganesh Del Grosso, Hamid Jalalzai, Georg Pichler, Catuscia Palamidessi, Pablo Piantanida
The use of personal data for training machine learning systems comes with a privacy threat and measuring the level of privacy of a model is one of the major challenges in machine learning today.
1 code implementation • ICLR 2022 • Eduardo Dadalto Camara Gomes, Florence Alberge, Pierre Duhamel, Pablo Piantanida
Reliable out-of-distribution (OOD) detection is fundamental to implementing safer modern machine learning (ML) systems.
Out-of-Distribution Detection
Out of Distribution (OOD) Detection
1 code implementation • 14 Feb 2022 • Georg Pichler, Pierre Colombo, Malik Boudiaf, Günther Koliander, Pablo Piantanida
Mutual Information (MI) has been widely used as a loss regularizer for training neural networks.
no code implementations • 10 Dec 2021 • Matias Vera, Leonardo Rey Vega, Pablo Piantanida
In this work, we introduce an analysis based on point-wise PAC approach over the generalization gap considering the mismatch of testing based on the accuracy metric and training on the negative log-loss.
1 code implementation • 2 Dec 2021 • Pierre Colombo, Chloe Clavel, Pablo Piantanida
In this paper, we introduce InfoLM a family of untrained metrics that can be viewed as a string-based metric that addresses the aforementioned flaws thanks to a pre-trained masked language model.
2 code implementations • EMNLP 2021 • Pierre Colombo, Guillaume Staerman, Chloe Clavel, Pablo Piantanida
A new metric \texttt{BaryScore} to evaluate text generation based on deep contextualized embeddings e. g., BERT, Roberta, ELMo) is introduced.
no code implementations • 17 Jul 2021 • Mohammadhadi Shateri, Francisco Messina, Pablo Piantanida, Fabrice Labeau
We formulate this as the problem of learning a sparse representation of SMs data with minimum information leakage and maximum utility.
no code implementations • 14 Jul 2021 • Carlos Pinzón, Catuscia Palamidessi, Pablo Piantanida, Frank Valencia
One of the main concerns about fairness in machine learning (ML) is that, in order to achieve it, one may have to trade off some accuracy.
3 code implementations • 23 Jun 2021 • Malik Boudiaf, Ziko Imtiaz Masud, Jérôme Rony, Jose Dolz, Ismail Ben Ayed, Pablo Piantanida
We motivate our transductive loss by deriving a formal relation between the classification accuracy and mutual-information maximization.
1 code implementation • 12 Jun 2021 • Marine Picot, Francisco Messina, Malik Boudiaf, Fabrice Labeau, Ismail Ben Ayed, Pablo Piantanida
Adversarial robustness has become a topic of growing interest in machine learning since it was observed that neural networks tend to be brittle.
1 code implementation • NeurIPS 2021 • Federica Granese, Marco Romanelli, Daniele Gorla, Catuscia Palamidessi, Pablo Piantanida
Deep neural networks (DNNs) have shown to perform very well on large scale object recognition problems and lead to widespread use for real-world applications, including situations where DNN are implemented as "black boxes".
no code implementations • 9 May 2021 • Ganesh Del Grosso, Georg Pichler, Catuscia Palamidessi, Pablo Piantanida
We present a novel formalism, generalizing membership and attribute inference attack setups previously studied in the literature and connecting them to memorization and generalization.
no code implementations • ACL 2021 • Pierre Colombo, Chloe Clavel, Pablo Piantanida
We show the superiority of this method on fair classification and on textual style transfer tasks.
Ranked #1 on
Text Style Transfer
on Yelp Review Dataset (Large)
(BLEU metric)
no code implementations • 1 Jan 2021 • Pierre Colombo, Chloé Clavel, Pablo Piantanida
Learning disentangled representations of textual data is essential for many natural language tasks such as fair classification (\textit{e. g.} building classifiers whose decisions cannot disproportionately hurt or benefit specific groups identified by sensitive attributes), style transfer and sentence generation, among others.
2 code implementations • CVPR 2021 • Malik Boudiaf, Hoel Kervadec, Ziko Imtiaz Masud, Pablo Piantanida, Ismail Ben Ayed, Jose Dolz
We show that the way inference is performed in few-shot segmentation tasks has a substantial effect on performances -- an aspect often overlooked in the literature in favor of the meta-learning paradigm.
Ranked #3 on
Few-Shot Semantic Segmentation
on COCO-20i (10-shot)
no code implementations • 6 Dec 2020 • Ganesh Del Grosso, Georg Pichler, Pablo Piantanida
However, the use of power consumption data raises significant privacy concerns, as this data usually belongs to clients of a power company.
1 code implementation • NeurIPS 2020 • Malik Boudiaf, Imtiaz Ziko, Jérôme Rony, Jose Dolz, Pablo Piantanida, Ismail Ben Ayed
We introduce Transductive Infomation Maximization (TIM) for few-shot learning.
no code implementations • 20 Nov 2020 • Mohammadhadi Shateri, Francisco Messina, Pablo Piantanida, Fabrice Labeau
In this paper, we study this problem in the context of time series data and smart meters (SMs) power consumption measurements in particular.
no code implementations • 22 Oct 2020 • Matias Vera, Leonardo Rey Vega, Pablo Piantanida
In practice, this behaviour is controlled by various--sometimes heuristics--regularization techniques, which are motivated by developing upper bounds to the generalization error.
2 code implementations • 25 Aug 2020 • Malik Boudiaf, Ziko Imtiaz Masud, Jérôme Rony, José Dolz, Pablo Piantanida, Ismail Ben Ayed
We introduce Transductive Infomation Maximization (TIM) for few-shot learning.
no code implementations • 29 Jun 2020 • Mohammadhadi Shateri, Francisco Messina, Pablo Piantanida, Fabrice Labeau
On the one hand, the releaser in the CAL method, by getting supervision from the actual values of the private variables and feedback from the adversary performance, tries to minimize the adversary log-likelihood.
no code implementations • 10 Jun 2020 • Mohammadhadi Shateri, Francisco Messina, Pablo Piantanida, Fabrice Labeau
Unlike previous studies, we model the whole temporal correlation in the data to learn the MI in its general form and use a neural network to estimate the MI-based reward signal to guide the PCMU learning process.
1 code implementation • 9 May 2020 • Marco Romanelli, Konstantinos Chatzikokolakis, Catuscia Palamidessi, Pablo Piantanida
A feature of our approach is that it does not require to estimate the conditional probabilities, and that it is suitable for a large class of ML algorithms.
1 code implementation • ECCV 2020 • Malik Boudiaf, Jérôme Rony, Imtiaz Masud Ziko, Eric Granger, Marco Pedersoli, Pablo Piantanida, Ismail Ben Ayed
Second, we show that, more generally, minimizing the cross-entropy is actually equivalent to maximizing the mutual information, to which we connect several well-known pairwise losses.
Ranked #12 on
Metric Learning
on CARS196
(using extra training data)
no code implementations • 10 Mar 2020 • Mohammadhadi Shateri, Francisco Messina, Pablo Piantanida, Fabrice Labeau
Smart meters (SMs) play a pivotal rule in the smart grid by being able to report the electricity usage of consumers to the utility provider (UP) almost in real-time.
no code implementations • 7 Feb 2020 • Georg Pichler, Pablo Piantanida, Günther Koliander
In particular, we provide confidence bounds for simple histogram based estimation of differential entropy from a fixed number of samples, assuming that the probability density function is Lipschitz continuous with known Lipschitz constant and known, bounded support.
no code implementations • 14 Jun 2019 • Mohammadhadi Shateri, Francisco Messina, Pablo Piantanida, Fabrice Labeau
In this paper, we focus on real-time privacy threats, i. e., potential attackers that try to infer sensitive information from SMs data in an online fashion.
no code implementations • 28 May 2019 • Matias Vera, Pablo Piantanida, Leonardo Rey Vega
Our main result is that the testing gap between the empirical cross-entropy and its statistical expectation (measured with respect to the testing probability law) can be bounded with high probability by the mutual information between the input testing samples and the corresponding representations, generated by the encoder obtained at training time.
no code implementations • MIDL 2019 • Georg Pichler, Jose Dolz, Ismail Ben Ayed, Pablo Piantanida
We juxtapose our approach to state-of-the-art segmentation adaptation via adversarial training in the network-output space.
1 code implementation • 26 Feb 2018 • Clément Feutry, Pablo Piantanida, Yoshua Bengio, Pierre Duhamel
Statistical methods protecting sensitive information or the identity of the data owner have become critical to ensure privacy of individuals as well as of organizations.
no code implementations • 14 Feb 2018 • Matías Vera, Pablo Piantanida, Leonardo Rey Vega
This paper presents a sample-dependent bound on the generalization gap of the cross-entropy loss that scales with the information complexity (IC) of the representations, meaning the mutual information between inputs and their representations.
no code implementations • 19 Nov 2017 • Matías Vera, Leonardo Rey Vega, Pablo Piantanida
This paper investigates, from information theoretic grounds, a learning problem based on the principle that any regularity in a given dataset can be exploited to extract compact features from data, i. e., using fewer bits than needed to fully describe the data itself, in order to build meaningful representations of a relevant content (multiple labels).
no code implementations • 14 Nov 2017 • Qianqian Yang, Pablo Piantanida, Deniz Gündüz
Based on information forwarded by the preceding layer, each stage of the network is required to preserve a certain level of relevance with regards to a specific hidden variable, quantified by the mutual information.
no code implementations • 5 Apr 2016 • Matías Vera, Leonardo Rey Vega, Pablo Piantanida
On the other hand, in CDIB there are two cooperating encoders which separately observe $X_1$ and $X_2$ and a third node which can listen to the exchanges between the two encoders in order to obtain information about a hidden variable $Y$.
no code implementations • 15 Feb 2016 • Georg Pichler, Pablo Piantanida, Gerald Matz
We study a novel multi-terminal source coding setup motivated by the biclustering problem.