6.1.1 CoRTeX

• Name: Python library for noun phrase COreference Resolution in natural language TEXts
• Keyword: Natural language processing
• Functional Description: CoRTex is a LGPL-licensed Python library for Noun Phrase coreference resolution in natural language texts. This library contains implementations of various state-of-the-art coreference resolution algorithms, including those developed in our research. In addition, it provides a set of APIs and utilities for text pre-processing, reading the CONLL2012 and CONLLU annotation formats, and performing evaluation, notably based on the main evaluation metrics (MUC, B-CUBED, and CEAF). As such, CoRTex provides benchmarks for researchers working on coreference resolution, but it is also of interest for developers who want to integrate a coreference resolution within a larger platform. It currently supports use of the English or French language.
• Contact: Pascal Denis
• Participant: Pascal Denis
• Partner: Orange Labs

6.1.2 Mangoes

• Name: MAgnet liNGuistic wOrd vEctorS
• Keywords: Word embeddings, NLP
• Functional Description:

  Mangoes is a toolbox for constructing and evaluating static and contextual token vector representations (aka embeddings). The main functionalities are:

  - Contextual embeddings: Access a large collection of pretrained transformer-based language models, Pre-train a BERT language model on a corpus, Fine-tune a BERT language model for a number of extrinsic tasks, Extract features/predictions from pretrained language models.

  - Static embeddings: Process textual data and compute vocabularies and co-occurrence matrices. Input data should be raw text or annotated text, Compute static word embeddings with different state-of-the art unsupervised methods, Propose statistical and intrinsic evaluation methods, as well as some visualization tools, Generate context dependent embeddings from a pretrained language model.

  Future releases will include methods for injecting lexical and semantic knowledge into token and multi-model embeddings, and interfaces into common external knowledge resources.

• URL: https://gitlab.inria.fr/magnet/mangoes
• Contact: Nathalie Vauquier

6.1.3 metric-learn

• Keywords: Machine learning, Python, Metric learning
• Functional Description:

  Distance metrics are widely used in the machine learning literature. Traditionally, practicioners would choose a standard distance metric (Euclidean, City-Block, Cosine, etc.) using a priori knowledge of the domain. Distance metric learning (or simply, metric learning) is the sub-field of machine learning dedicated to automatically constructing optimal distance metrics.

  This package contains efficient Python implementations of several popular metric learning algorithms.

• URL: https://github.com/scikit-learn-contrib/metric-learn
• Contact: Aurélien Bellet
• Partner: Parietal

6.1.4 MyLocalInfo

• Keywords: Privacy, Machine learning, Statistics
• Functional Description: Decentralized algorithms for machine learning and inference tasks which (1) perform as much computation as possible locally and (2) ensure privacy and security by avoiding that personal data leaves devices.
• Contact: Nathalie Vauquier

6.1.5 COMPRISE Voice Transformer

• Name: COMPRISE Voice Transformer
• Keywords: Speech, Privacy
• Functional Description: COMPRISE Voice Transformer is an open source tool that increases the privacy of users of voice interfaces by converting their voice into another person’s voice without modifying the spoken message. It ensures that any information extracted from the transformed voice can hardly be traced back to the original speaker, as validated through state-of-the-art biometric protocols, and it preserves the phonetic information required for human labelling and training of speech-to-text models.
• Release Contributions: This version gives access to the 2 generations of tools that have been used to transform the voice, as part of the COMPRISE project (https://www.compriseh2020.eu/). The first one is a python library that implements 2 basic voice conversion methods, both using VLTN. The second one implements an anonymization method using x-vectors and neural waveform models.
• URL: https://gitlab.inria.fr/comprise/voice_transformation
• Contact: Marc Tommasi
• Participants: Nathalie Vauquier, Brij Mohan Lal Srivastava, Marc Tommasi, Emmanuel Vincent, Md Sahidullah

Integrating knowledge graph embeddings to improve mention representation for bridging anaphora resolution

Lexical semantics and world knowledge are crucial for interpreting bridging anaphora1. Yet, existing computational methods for acquiring and injecting this type of information into bridging resolution systems suffer important limitations. Based on explicit querying of external knowledge bases, earlier approaches are computationally expensive (hence, hardly scalable) and they map the data to be processed into high-dimensional spaces (careful handling of the curse of dimensionality and overfitting has to be in order). In 24, we take a different and principled approach which naturally addresses these issues. Specifically, we convert the external knowledge source (in this case, WordNet) into a graph, and learn embeddings of the graph nodes of low dimension to capture the crucial features of the graph topology and, at the same time, rich semantic information. Once properly identified from the mention text spans, these low dimensional graph node embeddings are combined with distributional text-based embeddings to provide enhanced mention representations. We illustrate the effectiveness of our approach by evaluating it on commonly used datasets, namely ISNotes  36 and BASHI  37. Our enhanced mention representations yield significant accuracy improvements on both datasets when compared to different standalone text-based mention representations.

Fully Decentralized Joint Learning of Personalized Models and Collaboration Graphs

In 23, we consider the fully decentralized machine learning scenario where many users with personal datasets collaborate to learn models through local peer-to-peer exchanges, without a central coordinator. We propose to train personalized models that leverage a collaboration graph describing the relationships between user personal tasks, which we learn jointly with the models. Our fully decentralized optimization procedure alternates between training nonlinear models given the graph in a greedy boosting manner, and updating the collaboration graph (with controlled sparsity) given the models. Throughout the process, users exchange messages only with a small number of peers (their direct neighbors when updating the models, and a few random users when updating the graph), ensuring that the procedure naturally scales with the number of users. Overall, our approach is communication-efficient and avoids exchanging personal data. We provide an extensive analysis of the convergence rate, memory and communication complexity of our approach, and demonstrate its benefits compared to competing techniques on synthetic and real datasets.

Who started this rumor? Quantifying the natural differential privacy guarantees of gossip protocols

Gossip protocols are widely used to disseminate information in massive peer-to-peer networks. These protocols are often claimed to guarantee privacy because of the uncertainty they introduce on the node that started the dissemination. But is that claim really true? Can the source of a gossip safely hide in the crowd? In 1 we examine, for the first time, gossip protocols through a rigorous mathematical framework based on differential privacy to determine the extent to which the source of a gossip can be traceable. Considering the case of a complete graph in which a subset of the nodes are curious, we study a family of gossip protocols parameterized by a “muting” parameter $s$: nodes stop emitting after each communication with a fixed probability $1-s$. We first prove that the standard push protocol, corresponding to the case $s=1$, does not satisfy differential privacy for large graphs. In contrast, the protocol with $s=0$ achieves optimal privacy guarantees but at the cost of a drastic increase in the spreading time compared to standard push, revealing an interesting tension between privacy and spreading time. Yet, surprisingly, we show that some choices of the muting parameter $s$ lead to protocols that achieve an optimal order of magnitude in both privacy and speed. We also confirm empirically that, with appropriate choices of $s$, we indeed obtain protocols that are very robust against concrete source location attacks while spreading the information almost as fast as the standard (and non-private) push protocol.

Private Protocols for U-Statistics in the Local Model and Beyond

In 15, we study the problem of computing U-statistics of degree 2, i.e., quantities that come in the form of averages over pairs of data points, in the local model of differential privacy (LDP). The class of U-statistics covers many statistical estimates of interest, including Gini mean difference, Kendall's tau coefficient and Area under the ROC Curve (AUC), as well as empirical risk measures for machine learning problems such as ranking, clustering and metric learning. We first introduce an LDP protocol based on quantizing the data into bins and applying randomized response, which guarantees an $ϵ$-LDP estimate with a Mean Squared Error (MSE) of $O(1/\sqrt{n}ϵ)$ under regularity assumptions on the U-statistic or the data distribution. We then propose a specialized protocol for AUC based on a novel use of hierarchical histograms that achieves MSE of $O({\alpha }^3/n{ϵ}^2)$ for arbitrary data distribution on a domain with $2^{\alpha }$ elements. We also show that 2-party secure computation allows to design a protocol with MSE of $O(1/n{ϵ}^2)$, without any assumption on the kernel function or data distribution and with total communication linear in the number of users $n$. Finally, we evaluate the performance of our protocols through experiments on synthetic and real datasets.

Interpretable privacy with optimizable utility

In a position paper (25), we discuss the problem of specifying privacy requirements for machine learning based systems, in an interpretable yet operational way. Explaining privacy-improving technology is a challenging problem, especially when the goal is to construct a system which at the same time is interpretable and has a high performance. In order to address this challenge, we propose to specify privacy requirements as constraints, leaving several options for the concrete implementation of the system open, followed by a constraint optimization approach to achieve an efficient implementation also, next to the interpretable privacy guarantees.

Reconstructing Genotypes in Private Genomic Databases from Genetic Risk Scores

Some organisations like 23andMe and the UK Biobank have large genomic databases that they re-use for multiple different genomewide association studies (GWAS). Even research studies that compile smaller genomic databases often utilise these databases to investigate many related traits. It is common for the study to report a genetic risk score (GRS) model for each trait within the publication. In 18 we show that under some circumstances, these GRS models can be used to recover the genetic variants of individuals in these genomic databases—a reconstruction attack. In particular, if two GRS models are trained using a largely overlapping set of participants, then it is often possible to determine the genotype for each of the individuals who were used to train one GRS model, but not the other. We demonstrate this theoretically and experimentally by analysing the Cornell Dog Genome database. The accuracy of our reconstruction attack depends on how accurately we can estimate the rate of co-occurrence of pairs of single nucleotide polymorphisms (SNP) within the private database, so if this aggregate information is ever released, it would drastically reduce the security of a private genomic database. Caution should be applied when using the same database for multiple analysis, especially when a small number of individuals are included or excluded from one part of the study.

A Decentralized Framework for Biostatistics and Privacy Concerns

The paper 13 is the first result of the new collaboration engaged with the Lille Hospital (CHRU) and its INCLUDE team. Biostatistics and machine learning have been the cornerstone of a variety of recent developments in medicine. In order to gather large enough datasets, it is often necessary to set up multi-centric studies; yet, centralization of measurements can be difficult, either for practical, legal or ethical reasons. As an alternative, federated learning enables leveraging multiple centers' data without actually collating them. While existing works generally require a center to act as a leader and coordinate computations, we propose a fully decentralized framework where each center plays the same role. In this paper, we apply this framework to logistic regression, including confidence intervals computation. We test our algorithm on two distinct clinical datasets split among different centers, and show that it matches results from the centralized framework. In addition, we discuss possible privacy leaks and potential protection mechanisms, paving the way towards further research.

Distributed Differentially Private Averaging with Improved Utility and Robustness to Malicious Parties

Learning from data owned by several parties, as in federated learning, raises challenges regarding the privacy guarantees provided to participants and the correctness of the computation in the presence of malicious parties. In 32, 33, we tackle these challenges in the context of distributed averaging, an essential building block of distributed and federated learning. Our first contribution is a novel distributed differentially private protocol which can match the accuracy of the trusted curator model even when each party communicates with only a logarithmic number of other parties chosen at random. Our second contribution is to enable users to prove the correctness of their computations without compromising the efficiency and privacy guarantees of the protocol. Our construction relies on standard cryptographic primitives like commitment schemes and zero knowledge proofs.

Joint Learning of the Graph and the Data Representation for Graph-Based Semi-Supervised Learning

Graph-based semi-supervised learning is appealing when labels are scarce but large amounts of unlabeled data are available. These methods typically use a heuristic strategy to construct the graph based on some fixed data representation, independently of the available labels. In 22, we propose to jointly learn a data representation and a graph from both labeled and unlabeled data such that (i) the learned representation indirectly encodes the label information injected into the graph, and (ii) the graph provides a smooth topology with respect to the transformed data. Plugging the resulting graph and representation into existing graph-based semi-supervised learning algorithms like label spreading and graph convolutional networks, we show that our approach outperforms standard graph construction methods on both synthetic data and real datasets.

Skill Rating for Multiplayer Games Introducing Hypernode Graphs and their Spectral Theory

This result corresponds to an extended version 8 of the results obtained on learning with hypergraphs. We consider the skill rating problem for multiplayer games, that is how to infer player skills from game outcomes in multiplayer games. We formulate the problem as a minimization problem ${\mathrm{argmin}}_s{s}^{\top }\Delta s$ where $\Delta$ is a positive semidefinite matrix and $s$ a real-valued function, of which some entries are the skill values to be inferred and other entries are constrained by the game outcomes. We leverage graph-based semi-supervised learning (SSL) algorithms for this problem. We apply our algorithms on several data sets of multiplayer games and obtain very promising results compared to Elo Duelling  34 and TrueSkill  35. As we leverage graph-based SSL algorithms and because games can be seen as relations between sets of players, we then generalize the approach. For this aim, we introduce a new finite model, called hypernode graph, defined to be a set of weighted binary relations between sets of nodes. We define Laplacians of hypernode graphs. Then, we show that the skill rating problem for multiplayer games can be formulated as ${\mathrm{argmin}}_s{s}^{\top }\Delta s$ where $\Delta$ is the Laplacian of a hypernode graph constructed from a set of games. From a fundamental perspective, we show that hypernode graph Laplacians are symmetric positive semidefinite matrices with constant functions in their null space. We show that problems on hypernode graphs cannot be solved with graph constructions and graph kernels. We relate hypernode graphs to signed graphs showing that positive relations between groups can lead to negative relations between individuals.

metric-learn: Metric Learning Algorithms in Python

The paper 12 describes metric-learn, an open source Python package implementing supervised and weakly supervised distance metric learning algorithms. As part of scikit-learn-contrib, it provides a unified interface compatible with scikit-learn which allows to easily perform cross-validation, model selection, and pipelining with other machine learning estimators. metric-learn is thoroughly tested and available on PyPi under the MIT license.

Introducing the VoicePrivacy initiative

The VoicePrivacy initiative aims to promote the development of privacy preservation tools for speech technology by gathering a new community to define the tasks of interest and the evaluation methodology, and benchmarking solutions through a series of challenges. In 21, we formulate the voice anonymization task selected for the VoicePrivacy 2020 Challenge and describe the datasets used for system development and evaluation. We also present the attack models and the associated objective and subjective evaluation metrics. We introduce two anonymization baselines and report objective evaluation results.

Design Choices for X-vector Based Speaker Anonymization

The recently proposed x-vector based anonymization scheme converts any input voice into that of a random pseudo-speaker. In 19, we present a flexible pseudo-speaker selection technique as a baseline for the first VoicePrivacy Challenge. We explore several design choices for the distance metric between speakers, the region of x-vector space where the pseudo-speaker is picked, and gender selection. To assess the strength of anonymization achieved, we consider attackers using an x-vector based speaker verification system who may use original or anonymized speech for enrollment, depending on their knowledge of the anonymization scheme. The Equal Error Rate (EER) achieved by the attackers and the decoding Word Error Rate (WER) over anonymized data are reported as the measures of privacy and utility. Experiments are performed using datasets derived from LibriSpeech to find the optimal combination of design choices in terms of privacy and utility.

A comparative study of speech anonymization metrics

Speech anonymization techniques have recently been proposed for preserving speakers' privacy. They aim at concealing speak-ers' identities while preserving the spoken content. In 17, we compare three metrics proposed in the literature to assess the level of privacy achieved. We exhibit through simulation the differences and blindspots of some metrics. In addition, we conduct experiments on real data and state-of-the-art anonymiza-tion techniques to study how they behave in a practical scenario. We show that the application-independent log-likelihood-ratio cost function $C_{\mathrm{llr}}^{\min }$ provides a more robust evaluation of privacy than the equal error rate (EER), and that detection-based metrics provide different information from linkability metrics. Interestingly, the results on real data indicate that current anonymization design choices do not induce a regime where the differences between those metrics become apparent.

Evaluating Voice Conversion-based Privacy Protection against Informed Attackers

Speech data conveys sensitive speaker attributes like identity or accent. With a small amount of found data, such attributes can be inferred and exploited for malicious purposes: voice cloning, spoofing etc. Anonymization aims to make the data unlinkable, i.e., ensure that no utterance can be linked to its original speaker. In 9, we investigate anonymization methods based on voice conversion. In contrast to prior work, we argue that various linkage attacks can be designed depending on the attackers' knowledge about the anonymization scheme. We compare two frequency warping-based conversion methods and a deep learning based method in three attack scenarios. The utility of converted speech is measured via the word error rate achieved by automatic speech recognition, while privacy protection is assessed by the increase in equal error rate achieved by state-of-the-art i-vector or x-vector based speaker verification. Our results show that voice conversion schemes are unable to effectively protect against an attacker that has extensive knowledge of the type of conversion and how it has been applied, but may provide some protection against less knowledgeable attackers.

8.1.1 Inria International Labs

8.1.2 Inria associate team not involved in an IIL

8.1.3 Participation in other international programs

8.2.1 FP7 & H2020 Projects

8.2.2 Collaborations in European programs, except FP7 and H2020

ANR Pamela (2016–2022)

Participants: Marc Tommasi, Aurélien Bellet, Rémi Gilleron, Jan Ramon, Mahsa Asadi.

The Pamela project aims at developing machine learning theories and algorithms in order to learn local and personalized models from data distributed over networked infrastructures. Our project seeks to provide first answers to modern information systems built by interconnecting many personal devices holding private user data in the search of personalized suggestions and recommendations. More precisely, we will focus on learning in a collaborative way with the help of neighbors in a network. We aim to lay the first blocks of a scientific foundation for these new types of systems, in effect moving from graphs of data to graphs of data and learned models. We argue that this shift is necessary in order to address the new constraints arising from the decentralization of information that is inherent to the emergence of big data. We will in particular focus on the question of learning under communication and privacy constraints. A significant asset of the project is the quality of its industrial partners, Snips and Mediego, who bring in their expertise in privacy protection and distributed computing as well as use cases and datasets. They will contribute to translate this fundamental research effort into concrete outcomes by developing personalized and privacy-aware assistants able to provide contextualized recommendations on small devices and smartphones.

https://project.inria.fr/pamela/

ANR JCJC GRASP (2016–2022)

Participants: Pascal Denis, Aurélien Bellet, Rémi Gilleron, Mikaela Keller, Marc Tommasi.

The GRASP project aims at designing new graph-based Machine Learning algorithms that are better tailored to Natural Language Processing structured output problems. Focusing on semi-supervised learning scenarios, we will extend current graph-based learning approaches along two main directions: (i) the use of structured outputs during inference, and (ii) a graph construction mechanism that is more dependent on the task objective and more closely related to label inference. Combined, these two research strands will provide an important step towards delivering more adaptive (to new domains and languages), more accurate, and ultimately more useful language technologies. We will target semantic and pragmatic tasks such as coreference resolution, temporal chronology prediction, and discourse parsing for which proper Machine Learning solutions are still lacking.

https://project.inria.fr/grasp/

ANR DEEP-Privacy (2019–2023)

Participants: Marc Tommasi, Aurélien Bellet, Pascal Denis, Jan Ramon, Brij Mohan Lal Srivastava.

DEEP-PRIVACY proposes a new paradigm based on a distributed, personalized, and privacy-preserving approach for speech processing, with a focus on machine learning algorithms for speech recognition. To this end, we propose to rely on a hybrid approach: the device of each user does not share its raw speech data and runs some private computations locally, while some cross-user computations are done by communicating through a server (or a peer-to-peer network). To satisfy privacy requirements at the acoustic level, the information communicated to the server should not expose sensitive speaker information.

8.3.1 ANR-JCJC PRIDE (2020–2024)

Participants: Aurélien Bellet, Paul Mangold.

Machine learning (ML) is ubiquitous in AI-based services and data-oriented scientific fields but raises serious privacy concerns when training on personal data. The starting point of PRIDE is that personal data should belong to the individual who produces it. This requires to revisit ML algorithms to learn from many decentralized personal datasets while preventing the reconstruction of raw data. Differential Privacy (DP) provides a strong notion of protection, but current decentralized ML algorithms are not able to learn useful models under DP. The goal of PRIDE is to develop theoretical and algorithmic tools that enable differentially-private ML methods operating on decentralized datasets, through two complementary objectives: (1) prove that gossip protocols naturally reinforce DP guarantees; (2) propose algorithms at the intersection of decentralized ML and secure multi-party computation.

8.3.2 ANR PMR (2020-2024)

Participants: Jan Ramon.

Given the growing awareness of privacy risks of data processing, there is an increasing interest in privacy-preserving learning. However, shortcomings in the state of the art limit the applicability of the privacy-preserving learning paradigm. First, most approaches assume too optimistically a honest-but-curious setting. Second, most approaches consider one learning task in isolation, not accounting for the context where querying is a recurring activity. We will investigate new algorithms and models that address these shortcomings. Among others, (i) our algorithms will combine privacy-preserving properties of differential privacy with security offered by cryptography and (ii) based on models of information flows in integrated data handling processes, we will build more refined models analyzing the implications of repeated querying. We will demonstrate the utility of our new theory and algorithms by proposing strategies to realistically apply them in significant real-world problems illustrated through use cases in the medical domain

9.1.1 Scientific events: organisation

• Aurélien Bellet co-organized the Privacy Preserving Machine Learning (PPML'20) workshop at NeurIPS'20.2
• Since May 2020, Aurélien Bellet is co-organizing the Federated Learning One World webinar3 (700+ registered attendees)

9.1.2 Scientific events: selection

9.1.3 Invited talks

• Aurélien Bellet gave invited talks at the Federated Learning Winter School,4 the Google Workshop on Federated Learning and Analytics, the Applied Machine Learning Days,5 the CIRM Conference on Optimization for Machine Learning,6, the International Workshop on Distributed Cloud Computing7 and the International Workshop on Privacy and Personal Data Protection 8.
• Fabio Vitale gave an invited talk at ISI Foundation (Turin, Italy). Title: “Compressing graph information for binary node classification”. February, 2020.
• Fabio Vitale gave an invited talk at Technical University of Vienna (Austria). Title: “Fast Clustering through Pairwise Similarity Information”. September, 2020.

9.1.4 Scientific expertise

• Aurélien Bellet was a member of the jury for the Gilles-Kahn PhD award of the French Society of Computer Science (SIF), sponsored by the French Academy of Sciences.9
• Aurélien Bellet was a memberof the recruitment committee of an associate professor at Ecole Normale Supérieur de Lyon.
• Jan Ramon was a member of the jury for the European Young Researchers Award (EYRA) of Euroscience.
• Pascal Denis served as CoCNRS representative for the evaluation panel of Haut Conseil de l'évaluation de la recherche et de l'enseignement supérieur, HCERES10.
• Marc Tommasi was a member of the recruitment committee of full professors at Marseille and Saint-Etienne.
• Marc Tommasi served as an expert for ANRT
• Pascal Denis acted as vice-president (before resigning) of the Inria Lille Search committee for Junior Researchers CRCN/IFSP. He was also member of search committee for a MdC position at Université Paris-Diderot, Linguistics Department.

9.1.5 Research administration

• Marc Tommasi is co-head of the DatInG group (4 teams, about 100 persons), member of the Conseil Scientifique du laboratoire CRIStAL and member of the Commission mixte CRIStAL/Faculty of Science, Université de Lille.
• Marc Tommasi is Directeur d'Études for the Master of Computer Science, Opt. Machine Learning.
• Pascal Denis served as an elected member of the Comité National du CNRS, section 34 (Sciences du Langage).
• Pascal Denis is a member of the CNRS GDR NLP Group.
• Jan Ramon was a member of the INRIA-Lille committee emploi-recherche (CER).
• Aurélien Bellet is member of the Operational Committee for the assesment of Legal and Ethical risks (COERLE).
• Pascal Denis is administrator of Inria membership to Linguistic Data Consortium (LDC).

9.2.1 Teaching

• Licence SCE: Fabio Vitale, Apprentissage et émergence des comportements, 30h, L2, Université de Lille.
• Licence MIASHS: Fabio Vitale, Algorithmique et programmation, 42h, L3, Université de Lille.
• Licence SoQ: Fabio Vitale, Bases de données et SQL, 16h, L1, Université de Lille.
• Licence MIASHS: Fabio Vitale, Introduction aux bases de données, 18h, L3, Université de Lille.
• Licence SHS: Marc Tommasi, Data Science, 24h, L2, Université de Lille.
• Licence MIASHS: Marc Tommasi, Python Programming, 48h, L2 Université de Lille.
• Licence MIASHS: Marc Tommasi, Data Science, 24h, L2 Université de Lille.
• Licence MIASHS: Mikaela Keller, Python II, 40h, L2, Université de Lille.
• Licence MIASHS: Mikaela Keller, Traitement de données, 42h, L2, Université de Lille.
• Licence SoQ (SHS): Mikaela Keller, Algorithmique de graphes, 24h, L3, Université de Lille.
• Master MIASHS: Marc Tommasi, Python Programming, 45h, M1 Université de Lille.
• Master MIASHS: Mikaela Keller, Algorithmes fondamentaux de la fouille de données, 60h, M1, Université de Lille.
• Master Computer Science: Marc Tommasi, Machine Learning, 48h, M1, Université de Lille.
• Master Data Science: Marc Tommasi Seminars 24h.
• Master Data Science: Marc Tommasi Reading groups 24h.
• Master Data Science: Fabio Vitale, Algorithms and their complexity, 30h, M1, Ecole Centrale de Lille.
• Master Data Science: Aurélien Bellet, Privacy Preserving Machine Learning, 24h, M2, Université de Lille and Ecole Centrale de Lille.
• Master Data Analysis & Decision Making: Aurélien Bellet, Machine Learning, 12h, Ecole Centrale de Lille.
• Master / Master Spécialisé Big Data: Aurélien Bellet, Advanced Machine Learning, 15h, Télécom ParisTech.
• Formation continue (Certificat d’Études Spécialisées Data Scientist): Aurélien Bellet, Supervised Learning and Support Vector Machines, 14h, Télécom ParisTech.
• Master Informatique: Pascal Denis, Foundations of Machine Learning, 46h, M1, Université de Lille.
• Master SCE: Michaël Perrot, Machine Learning for Cognitive Sciences, 30h, M2, Université de Lille.

9.2.2 Supervision

• Postdoc: Mohamed Maouche, supervised by Aurélien Bellet, Marc Tommasi, Privacy attacks on representation learning for speech processing, since November 2019.
• PhD in progress: Onkar Pandit, Graph-based Semi-supervised Linguistic Structure Prediction, since Dec. 2017, Pascal Denis, Marc Tommasi and Liva Ralaivola (Aix-Marseille Université).
• PhD in progress: Mariana Vargas Vieyra, Adaptive Graph Learning with Applications to Natural Language Processing, since Jan. 2018. Pascal Denis and Aurélien Bellet and Marc Tommasi.
• PhD in progress: Brij Mohan Lal Srivastava, Representation Learning for Privacy-Preserving Speech Recognition, since Oct 2018 Aurélien Bellet and Marc Tommasi and Emmanuel Vincent.
• PhD in progress: Mahsa Asadi, On Decentralized Machine Learning, since Oct 2018. Aurélien Bellet and Marc Tommasi.
• PhD in progress: Nicolas Crosetti, Privacy Risks of Aggregates in Data Centric-Workflows, since Oct 2018. Florent Capelli and Sophie Tison and Joachim Niehren and Jan Ramon.
• PhD Robin Vogel, Learning to rank by similarity and performance optimization in biometric identification, 2017-2020 (CIFRE thesis with IDEMIA and Télécom ParisTech). Aurélien Bellet, Stéphan Clémençon and Anne Sabourin.
• PhD in progress: Moitree Basu, Integrated privacy-preserving AI, since 2019. Jan Ramon.
• PhD in progress: César Sabater, Privacy Preserving Machine Learning, since 2019. Jan Ramon.
• PhD in progress: Paul Mangold, supervised by Aurélien Bellet and Marc Tommasi and Joseph Salmon, since October 2020.
• Engineer Yannick Bouillard, supervised by Aurélien Bellet since Nov. 2020.
• Engineer Sophie Villerot, supervised by Jan Ramon since Nov. 2020.

9.2.3 Juries

• Aurélien Bellet was member of the PhD jury of Kevin Elgui (Télécom Paris)
• Jan Ramon was member of the PhD jury of Robin Vandaele (Ghent, BE)
• Marc Tommasi was member of the PhD jury of Antoine Chatalic (Rennes Univ.), Maziar Moradi Fard (Grenoble Univ.), Léo Gautheron (Saint-Etienne Univ.)
• Mikaela Keller was member of the PhD jury of Timothée Lacroix (Paris-Est Univ.)

International journals

• 12 articleWilliamW. De Vazelhes, CJC. Carey, YuanY. Tang, NathalieN. Vauquier and AurélienA. Bellet. metric-learn: Metric Learning Algorithms in PythonJournal of Machine Learning Research212020, 1-6
  HALback to text
• 13 article PaulP. Mangold, AlexandreA. Filiot, MouhamedM. Moussa, VincentV. Sobanski, GrégoireG. Ficheur, PaulP. Andrey and AntoineA. Lamer. A Decentralized Framework for Biostatistics and Privacy Concerns Studies in Health Technology and Informatics November 2020
  HAL DOI back to text
• 14 articleThomasT. Ricatte, RémiR. Gilleron and MarcM. Tommasi. Skill Rating for Multiplayer Games Introducing Hypernode Graphs and their Spectral TheoryJournal of Machine Learning Research212020, 1 - 18
  HAL

International peer-reviewed conferences

• 15 inproceedings JamesJ. Bell, AurélienA. Bellet, AdriàA. Gascón and TejasT. Kulkarni. Private Protocols for U-Statistics in the Local Model and Beyond AISTATS 2020 - 23rd International Conference on Artificial Intelligence and Statistics Palermo, Italy August 2020
  HAL back to text
• 16 inproceedings AurélienA. Bellet, RachidR. Guerraoui and HadrienH. Hendrikx. Who started this rumor? Quantifying the natural differential privacy guarantees of gossip protocols DISC 2020 - 34th International Symposium on Distributed Computing Freiburg / Virtual, Germany Inria 2020
  HAL
• 17 inproceedings MohamedM. Maouche, Brij Mohan LalB. Srivastava, NathalieN. Vauquier, AurélienA. Bellet, MarcM. Tommasi and EmmanuelE. Vincent. A comparative study of speech anonymization metrics INTERSPEECH 2020 Shanghai, China October 2020
  HAL back to text
• 18 inproceedings BrooksB. Paige, JamesJ. Bell, AurélienA. Bellet, AdriàA. Gascón and DaphneD. Ezer. Reconstructing Genotypes in Private Genomic Databases from Genetic Risk Scores 24th International Conference On Research In Computational Molecular Biology (RECOMB 2020) Virtual, Italy 2020
  HAL DOI back to text
• 19 inproceedings Brij Mohan LalB. Srivastava, NataliaN. Tomashenko, XinX. Wang, EmmanuelE. Vincent, JunichiJ. Yamagishi, MohamedM. Maouche, AurélienA. Bellet and MarcM. Tommasi. Design Choices for X-vector Based Speaker Anonymization INTERSPEECH 2020 Shanghai, China October 2020
  HAL back to text
• 20 inproceedingsBrij Mohan LalB. Srivastava, NathalieN. Vauquier, MdM. Sahidullah, AurélienA. Bellet, MarcM. Tommasi and EmmanuelE. Vincent. Evaluating Voice Conversion-based Privacy Protection against Informed AttackersICASSP 2020 - 45th International Conference on Acoustics, Speech, and Signal ProcessingBarcelona, SpainMay 2020, 2802-2806
  HAL
• 21 inproceedings NataliaN. Tomashenko, Brij Mohan LalB. Srivastava, XinX. Wang, EmmanuelE. Vincent, AndreasA. Nautsch, JunichiJ. Yamagishi, NicholasN. Evans, JoseJ. Patino, Jean-FrançoisJ.-F. Bonastre, Paul-GauthierP.-G. Noé and MassimilianoM. Todisco. Introducing the VoicePrivacy initiative INTERSPEECH 2020 Shanghai, China October 2020
  HAL back to text
• 22 inproceedings MarianaM. Vargas-Vieyra, AurélienA. Bellet and PascalP. Denis. Joint Learning of the Graph and the Data Representation for Graph-Based Semi-Supervised Learning 14th Workshop on Graph-Based Natural Language Processing (TextGraphs 2020) Virtual, Spain 2020
  HAL back to text
• 23 inproceedings ValentinaV. Zantedeschi, AurélienA. Bellet and MarcM. Tommasi. Fully Decentralized Joint Learning of Personalized Models and Collaboration Graphs AISTATS 2020 - The 23rd International Conference on Artificial Intelligence and Statistics Palerme / Virtual, Italy August 2020
  HAL back to text

Conferences without proceedings

• 24 inproceedings OnkarO. Pandit, PascalP. Denis and LivaL. Ralaivola. Integrating knowledge graph embeddings to improve mention representation for bridging anaphora resolution CRAC 2020 - Third Workshop on Computational Models of Reference, Anaphora and Coreference Virtual, France December 2020
  HAL back to text
• 25 inproceedings JanJ. Ramon and MoitreeM. Basu. Interpretable privacy with optimizable utility ECML/PKDD 2020 - Workshop on eXplainable Knowledge Discovery in Data mining Ghent / Virtual, Belgium September 2020
  HAL back to text

Scientific book chapters

• 26 inbookOlivierO. Bournez, GillesG. Dowek, RémiR. Gilleron, SergeS. Grigorieff, Jean-YvesJ.-Y. Marion, SimonS. Perdrix and S. Tison. Theoretical Computer Science: Computability, Decidability and LogicA Guided Tour of Artificial Intelligence Research - Volume III: Interfaces and Applications of Artificial Intelligence (10.1007/978-3-030-06170-8)Springer International Publishing2020, 1-50
  HALDOI
• 27 inbook OlivierO. Bournez, GillesG. Dowek, RémiR. Gilleron, SergeS. Grigorieff, Jean-YvesJ.-Y. Marion, SimonS. Perdrix and S. Tison. Theoretical Computer Science: Computational Complexity A Guided Tour of Artificial Intelligence Research - Volume III: Interfaces and Applications of Artificial Intelligence (10.1007/978-3-030-06170-8) Springer International Publishing 2020
  HAL

Reports & preprints

• 28 misc AurélienA. Bellet, PascalP. Denis, RémiR. Gilleron, MikaelaM. Keller and NathalieN. Vauquier. For more transparency in the automatic analysis of public consultations: lessons learned from the French "Grand Débat National" October 2020
  HAL
• 29 misc EdwigeE. Cyffers and AurélienA. Bellet. Privacy Amplification by Decentralization 2020
  HAL
• 30 misc Brij Mohan LalB. Srivastava, MohamedM. Maouche, MdM. Sahidullah, EmmanuelE. Vincent, AurélienA. Bellet, MarcM. Tommasi, NataliaN. Tomashenko, XinX. Wang and JunichiJ. Yamagishi. Privacy and utility of x-vector based speaker anonymization April 2021
  HAL
• 31 misc RobinR. Vogel, AurélienA. Bellet and StéphanS. Clémençon. Learning Fair Scoring Functions: Bipartite Ranking under ROC-based Fairness Constraints 2020
  HAL

Other scientific publications

• 32 misc CésarC. Sabater, AurélienA. Bellet and JanJ. Ramon. Distributed Differentially Private Averaging with Improved Utility and Robustness to Malicious Parties Vancouver (Virtual Workshop), Canada December 2020
  HAL back to text
• 33 misc CésarC. Sabater, AurélienA. Bellet and JanJ. Ramon. Échange de bruit corrélé pour le calcul distribué de moyenne avec garanties de confidentialité différentielle Vannes (Virtual), France June 2020
  HAL back to text