Publications

2021

• Learning from evolving data streams through ensembles of random patches
  
  • Gomes Heitor Murilo
  • Read Jesse
  • Bifet Albert
  • Durrant Robert J.
  Knowledge and Information Systems (KAIS), Springer, 2021, 63 (7), pp.1597--1625. Ensemble methods represent an effective way to solve supervised learning problems. Such methods are prevalent for learning from evolving data streams. One of the main reasons for such popularity is the possibility of incorporating concept drift detection and recovery strategies in conjunction with the ensemble algorithm. On top of that, successful ensemble strategies, such as bagging and random forest, can be easily adapted to a streaming setting. In this work, we analyse a novel ensemble method designed specially to cope with evolving data streams, namely the streaming random patches (SRP) algorithm. SRP combines random subspaces and online bagging to achieve competitive predictive performance in comparison with other methods. We significantly extend previous theoretical insights and empirical results illustrating different aspects of SRP. In particular, we explain how the widely adopted incremental Hoeffding trees are not, in fact, unstable learners, unlike their batch counterparts, and how this fact significantly influences ensemble methods design and performance. We compare SRP against state-of-the-art ensemble variants for streaming data in a multitude of datasets. The results show how SRP produces a high predictive performance for both real and synthetic datasets. We also show how ensembles of random subspaces can be an efficient and accurate option to SRP and leveraging bagging as we increase the number of base learners. Besides, we analyse the diversity over time and the average tree depth, which provides insights on the differences between local subspace randomization (as in random forest) and global subspace randomization (as in random subspaces). Finally, we analyse the behaviour of SRP when using Naive Bayes as its base learner instead of Hoeffding trees. (10.1007/S10115-021-01579-Z)
  DOI : 10.1007/S10115-021-01579-Z
• Proving the Safety of a Sliding Window Protocol with Event-B
  
  • Coudert Sophie
  , 2021.
• Multi-paradigm modelling for cyber–physical systems: a descriptive framework
  
  • Amrani Moussa
  • Blouin Dominique
  • Heinrich Robert
  • Rensink Arend
  • Vangheluwe Hans
  • Wortmann Andreas
  Software and Systems Modeling, Springer Verlag, 2021, 20, pp.611 - 639. The complexity of cyber-physical systems (CPSs) is commonly addressed through complex workflows, involving models in a plethora of different formalisms, each with their own methods, techniques, and tools. Some workflow patterns, combined with particular types of formalisms and operations on models in these formalisms, are used successfully in engineering practice. To identify and reuse them, we refer to these combinations of workflow and formalism patterns as modelling paradigms. This paper proposes a unifying (Descriptive) Framework to describe these paradigms, as well as their combinations. This work is set in the context of Multi-Paradigm Modelling (MPM), which is based on the principle to model every part and aspect of a system explicitly, at the most appropriate level(s) of abstraction, using the most appropriate modelling formalism(s) and workflows. The purpose of the Descriptive Framework presented in this paper is to serve as a basis to reason about these formalisms, workflows, and their combinations. One crucial part of the framework is the ability to capture the structural essence of a paradigm through the concept of a paradigmatic structure. This is illustrated informally by means of two example paradigms commonly used in CPS: Discrete Event Dynamic Systems and Synchronous Data Flow. The presented framework also identifies the need to establish whether a paradigm candidate follows, or qualifies as, a (given) paradigm. To illustrate the ability of the framework to support combining paradigms, the paper shows examples of both workflow and formalism combinations. The presented framework is intended as a basis for characterisation and classification of paradigms, as a starting point for a rigorous formalisation of the framework (allowing formal analyses), and as a foundation for MPM tool development. (10.1007/s10270-021-00876-z)
  DOI : 10.1007/s10270-021-00876-z
• Spatially relaxed inference on high-dimensional linear models
  
  • Chevalier Jérôme-Alexis
  • Nguyen Tuan-Binh
  • Thirion Bertrand
  • Salmon Joseph
  Statistics and Computing, Springer Verlag (Germany), 2021, 32 (83). We consider the inference problem for high-dimensional linear models, when covariates have an underlying spatial organization reflected in their correlation. A typical example of such a setting is high-resolution imaging, in which neighboring pixels are usually very similar. Accurate point and confidence intervals estimation is not possible in this context with many more covariates than samples, furthermore with high correlation between covariates. This calls for a reformulation of the statistical inference problem, that takes into account the underlying spatial structure: if covariates are locally correlated, it is acceptable to detect them up to a given spatial uncertainty. We thus propose to rely on the δ-FWER, that is the probability of making a false discovery at a distance greater than δ from any true positive. With this target measure in mind, we study the properties of ensembled clustered inference algorithms which combine three techniques: spatially constrained clustering, statistical inference, and ensembling to aggregate several clustered inference solutions. We show that ensembled clustered inference algorithms control the δ-FWER under standard assumptions for δ equal to the largest cluster diameter. We complement the theoretical analysis with empirical results, demonstrating accurate δ-FWER control and decent power achieved by such inference algorithms. (10.1007/s11222-022-10139-6)
  DOI : 10.1007/s11222-022-10139-6
• Nested Learning for Multi-Level Classification
  
  • Achddou Raphaël
  • Di Martino J. Matias
  • Sapiro Guillermo
  , 2021. Deep neural networks models are generally designed and trained for a specific type and quality of data. In this work, we address this problem in the context of nested learning. For many applications, both the input data, at training and testing, and the prediction can be conceived at multiple nested quality/resolutions. We show that by leveraging this multiscale information, the problem of poor generalization and prediction overconfidence, as well as the exploitation of multiple training data quality, can be efficiently addressed. We evaluate the proposed ideas in six public datasets: MNIST, Fashion-MNIST, CIFAR10, CIFAR100, Plantvillage, and DBPEDIA. We observe that coarsely annotated data can help to solve fine predictions and reduce overconfidence significantly. We also show that hierarchical learning produces models intrinsically more robust to adversarial attacks and data perturbations.
• Self-Supervised VQ-VAE for One-Shot Music Style Transfer
  
  • Cífka Ondřej
  • Ozerov Alexey
  • Şimşekli Umut
  • Richard Gael
  , 2021. Neural style transfer, allowing to apply the artistic style of one image to another, has become one of the most widely showcased computer vision applications shortly after its introduction. In contrast, related tasks in the music audio domain remained, until recently, largely untackled. While several style conversion methods tailored to musical signals have been proposed, most lack the 'one-shot' capability of classical image style transfer algorithms. On the other hand, the results of existing one-shot audio style transfer methods on musical inputs are not as compelling. In this work, we are specifically interested in the problem of one-shot timbre transfer. We present a novel method for this task, based on an extension of the vector-quantized variational autoencoder (VQ-VAE), along with a simple self-supervised learning strategy designed to obtain disentangled representations of timbre and pitch. We evaluate the method using a set of objective metrics and show that it is able to outperform selected baselines. (10.1109/ICASSP39728.2021.9414235)
  DOI : 10.1109/ICASSP39728.2021.9414235
• Neural Knowledge Base Repairs
  
  • Pellissier Tanon Thomas
  • Suchanek Fabian M.
  , 2021, pp.287-303. The curation of a knowledge base is a crucial but costly task. In this work, we suggest to make use of the advances in neural network research to improve the automated correction of constraint violations. Our method is a deep learning refinement of "Learning how to correct a knowledge base from the edit history", and similarly uses the edits that solved some violations in the past to infer how to solve similar violations in the present. Our system makes use of the graph content, literal embeddings, and features extracted from Web pages to improve its performance. The experimental evaluation on Wikidata shows significant improvements over baselines. (10.1007/978-3-030-77385-4_17)
  DOI : 10.1007/978-3-030-77385-4_17
• NEURO-STEERED MUSIC SOURCE SEPARATION WITH EEG-BASED AUDITORY ATTENTION DECODING AND CONTRASTIVE-NMF
  
  • Cantisani Giorgia
  • Essid Slim
  • Richard Gael
  , 2021. We propose a novel informed music source separation paradigm, which can be referred to as neuro-steered music source separation. More precisely, the source separation process is guided by the user's selective auditory attention decoded from his/her EEG response to the stimulus. This high-level prior information is used to select the desired instrument to isolate and to adapt the generic source separation model to the observed signal. To this aim, we leverage the fact that the attended instrument's neural encoding is substantially stronger than the one of the unattended sources left in the mixture. This "contrast" is extracted using an attention decoder and used to inform a source separation model based on non-negative matrix fac-torization named Contrastive-NMF. The results are promising and show that the EEG information can automatically select the desired source to enhance and improve the separation quality. (10.1109/ICASSP39728.2021.9413841)
  DOI : 10.1109/ICASSP39728.2021.9413841
• Distributed speech separation in spatially unconstrained microphone arrays
  
  • Furnon Nicolas
  • Serizel Romain
  • Illina Irina
  • Essid Slim
  , 2021. Speech separation with several speakers is a challenging task because of the non-stationarity of the speech and the strong signal similarity between interferent sources. Current state-of-the-art solutions can separate well the different sources using sophisticated deep neural networks which are very tedious to train. When several microphones are available, spatial information can be exploited to design much simpler algorithms to discriminate speakers. We propose a distributed algorithm that can process spatial information in a spatially unconstrained microphone array. The algorithm relies on a convolutional recurrent neural network that can exploit the signal diversity from the distributed nodes. In a typical case of a meeting room, this algorithm can capture an estimate of each source in a first step and propagate it over the microphone array in order to increase the separation performance in a second step. We show that this approach performs even better when the number of sources and nodes increases. We also study the influence of a mismatch in the number of sources between the training and testing conditions. (10.1109/ICASSP39728.2021.9414758)
  DOI : 10.1109/ICASSP39728.2021.9414758
• An OFDM-MIMO Distributed Acoustic Sensing over Deployed Telecom Fibers
  
  • Dorize Christian
  • Guerrier Sterenn
  • Awwad Elie
  • Nwakamma Peter A
  • Mardoyan Haik
  • Renaudier Jérémie
  , 2021. We demonstrate the performance of a novel multi-carrier MIMO fiber sensing interrogator based on OFDM approach. Over an installed telecom fiber cable, coherent fading is mitigated by means of a purely digital subcarrier combiner.
• PATCH DECODER-SIDE DEPTH ESTIMATION IN MPEG IMMERSIVE VIDEO
  
  • Milovanovic Marta
  • Henry Felix
  • Cagnazzo Marco
  • Jung Joel
  , 2021, pp.1945-1949. This paper presents a new approach for achieving bitrate and pixel rate reduction in the MPEG immersive video coding setting. We demonstrate that it is possible to avoid the transmission of some depth information in the Test Model for Immersive Video (TMIV) by estimating it at the receiver's side. Although the transmitted information in TMIV is considered as non-redundant, we show that it is possible to improve this algorithm. This method provides 3.4%, 9.0%, and 12.1% average BD-rate gain for natural content on high, medium, and low bitrate, respectively, with up to respectively 12.3%, 16.0%, and 18.4% peak reductions. Moreover, it preserves the perceptual quality as measured with MS-SSIM and VMAF metrics. Additionally, it decreases the pixel rate by 8.3% for each test sequence. (10.1109/ICASSP39728.2021.9414056)
  DOI : 10.1109/ICASSP39728.2021.9414056
• Ultra-low bitrate video conferencing using deep image animation
  
  • Konuko Goluck
  • Valenzise Giuseppe
  • Lathuilière Stéphane
  , 2021. In this work we propose a novel deep learning approach for ultra-low bitrate video compression for video conferencing applications. To address the shortcomings of current video compression paradigms when the available bandwidth is extremely limited, we adopt a model-based approach that employs deep neural networks to encode motion information as keypoint displacement and reconstruct the video signal at the decoder side. The overall system is trained in an end-to-end fashion minimizing a reconstruction error on the encoder output. Objective and subjective quality evaluation experiments demonstrate that the proposed approach provides an average bitrate reduction for the same visual quality of more than 80% compared to HEVC. (10.1109/icassp39728.2021.9414731)
  DOI : 10.1109/icassp39728.2021.9414731
• System for measurement of impedance cardiography
  
  • Maldari Mirko
  , 2021.
• Benefits of Local Cooperation in Sectorized Cellular Networks under a Complexity Constraint
  
  • Gelincik Samet
  • Wigger Michele
  • Wang Ligong
  IEEE Transactions on Wireless Communications, Institute of Electrical and Electronics Engineers, 2021, 20 (6), pp.3897-3910. The paper presents upper and lower bounds on the degrees of freedom (DoF) of a sectorized hexagonal cellular model when neighboring base stations (BSs) can cooperate during at most κ interaction rounds over rate-limited backhaul links. The lower bound is based on practically implementable beamforming and adapts the way BSs cooperate to the sectorization of the cells. It improves over the naive approach that ignores this sectorization in terms of the sum-rate, both at finite signal-to-noise ratio (SNR) and in the high-SNR limit. For moderate SNR, the new scheme improves also over an opportunistic cooperation strategy where each message is decoded based on the signals received at the three adjacent sectors with the best SNR. The upper bound is information-theoretic and holds for all possible coding schemes, including for example ergodic interference alignment whose practical implementation currently seems out of reach. Lower and upper bounds show that the complexity constraint, imposed by limiting the number of interaction rounds κ, indeed limits the largest achievable sum-rate and DoF. In particular, irrespective of the backhaul capacity µ, the per-user DoF cannot exceed a threshold which depends on κ. (10.1109/TWC.2021.3054337)
  DOI : 10.1109/TWC.2021.3054337
• Le low-tech et les réseaux : une rencontre impossible ?
  
  • Morgand Eva
  • Emmanuelle Frenoux
  • Coupechoux Marceau
  , 2021.
• Guest Editorial Special Issue: “From Deletion-Correction to Graph Reconstruction: In Memory of Vladimir I. Levenshtein”
  
  • Barg Alexander
  • Dolecek Lara
  • Gabrys Ryan
  • Katona Gyula
  • Korner Janos
  • Mcgregor Andrew
  • Milenkovic Olgica
  • Mesnager Sihem
  • Zemor Gilles
  IEEE Transactions on Information Theory, Institute of Electrical and Electronics Engineers, 2021, 67 (6), pp.3187-3189. (10.1109/TIT.2021.3072555)
  DOI : 10.1109/TIT.2021.3072555
• Data to Physicalization: A Survey of the Physical Rendering Process
  
  • Djavaherpour Hessam
  • Samavati Faramarz
  • Mahdavi-Amiri Ali
  • Yazdanbakhsh Fatemeh
  • Huron Samuel
  • Levy Richard
  • Jansen Yvonne
  • Oehlberg Lora
  Computer Graphics Forum, Wiley, 2021, 40 (3), pp.569-598. Physical representations of data offer physical and spatial ways of looking at, navigating, and interacting with data. While digital fabrication has facilitated the creation of objects with data-driven geometry, rendering data as a physically fabricated object is still a daunting leap for many physicalization designers. Rendering in the scope of this research refers to the back-and-forth process from digital design to digital fabrication and its specific challenges. We developed a corpus of example data physicalizations from research literature and physicalization practice. This survey then unpacks the "rendering" phase of the extended InfoVis pipeline in greater detail through these examples, with the aim of identifying ways that researchers, artists, and industry practitioners "render" physicalizations using digital design and fabrication tools. (10.1111/cgf.14330)
  DOI : 10.1111/cgf.14330
• Effect of Shockley-Read-Hall recombination on the static and dynamical characteristics of epitaxial quantum-dot lasers on silicon
  
  • Zhao Shiyuan
  • Grillot Frédéric
  Physical Review A, American Physical Society, 2021, 103 (6). (10.1103/PhysRevA.103.063521)
  DOI : 10.1103/PhysRevA.103.063521
• Digital Predistortion: principles, techniques and trends
  
  • Pham Dang-Kièn Germain
  • Desgreys Patricia
  , 2021.
• Linear codes from vectorial Boolean functions in the context of algebraic attacks
  
  • Boumezbeur M.
  • Mesnager Sihem
  • Guenda K.
  Discrete Mathematics, Algorithms and Applications, World Scientific Publishing, 2021, 13 (03), pp.2150032. In this paper, we study the relationship between vectorial (Boolean) functions and cyclic codes in the context of algebraic attacks. We first derive a direct link between the annihilators of a vectorial function (in univariate form) and certain [Formula: see text]-ary cyclic codes (which we show that they are LCD codes). We also present some properties of those cyclic codes as well as their weight enumerator. In addition, we generalize the so-called algebraic complement and study its properties. (10.1142/S1793830921500324)
  DOI : 10.1142/S1793830921500324
• Non-named Entities – The Silent Majority
  
  • Paris Pierre-Henri
  • Suchanek Fabian
  , 2021, 12739, pp.131-135. (10.1007/978-3-030-80418-3_24)
  DOI : 10.1007/978-3-030-80418-3_24
• T-P Thalès-Pythagore
  
  • Zayana Karim
  • Jacquet-Malo Lucie
  • Rabiet Victor
  CultureMath, ENS, 2021. Avec cet article, les auteurs nous proposent des Travaux Pratiques autour (des théorèmes) de Thalès et de Pythagore. Dans l'inconscient collectif, ces deux noms bien connus des collégiens symbolisent presque à eux seuls la géométrie-du moins dans un cadre scolaire. Or, que peut-on dire du rapport entre les deux ? Dans le cas général, ils ne coexistent pas toujours : le théorème de Pythagore exige un produit scalaire, tandis que le théorème de Thalès se contente d'un espace affine 1. L'intérêt est ici, dans le cas particulier du bon vieux plan euclidien, de les faire dialoguer, jusqu'à la résonance. Matériel utilisé : règle, compas, équerre, rapporteur. Notions abordées : angles correspondants, angles inscrits, arc capable, triangle isocèle, théorème de Thalès, théorème de Pythagore, cocyclicité, puissance d'un point par rapport à un cercle, symétrie, second degré.
• Blind Neural Belief Propagation Decoder for Linear Block Codes
  
  • Larue Guillaume
  • Dufrene Louis-Adrien
  • Lampin Quentin
  • Chollet Paul
  • Ghauch Hadi
  • Rekaya Ghaya
  , 2021. Neural belief propagation decoders were recently introduced by Nachmani et al. as a way to improve the decoding performance of belief propagation iterative algorithm for short to medium length linear block codes. The main idea behind these decoders is to represent belief propagation as a neural network, enabling adaptive weighting of the decoding process. In the present paper an efficient recurrent neural network architecture, based on gating and weights sharing mechanisms, is proposed to perform blind neural belief propagation decoding without prior knowledge of the coding scheme used by the encoder. The proposed architecture is able to learn to decode BCH (15,11) and BCH (15,7) codes and significantly improves the decoding performance over a standard belief propagation algorithm. A particular emphasis is given to the interpretability and complexity of the proposed model to ensure scalability to larger codes.
• La tête à Toto
  
  • Zayana Karim
  • Rabiet Victor
  CultureMath, ENS, 2021.
• Cyclic Bent Functions and Their Applications in Sequences
  
  • Abdukhalikov Kanat
  • Ding Cunsheng
  • Mesnager Sihem
  • Tang Chunming
  • Xiong Maosheng
  IEEE Transactions on Information Theory, Institute of Electrical and Electronics Engineers, 2021, 67 (6), pp.3473-3485. (10.1109/TIT.2021.3057896)
  DOI : 10.1109/TIT.2021.3057896