Publications

2021

• Feature Clustering for Support Identification in Extreme Regions
  
  • Jalalzai Hamid
  • Leluc Rémi
  Proceedings of Machine Learning Research, PMLR, 2021, 139, pp.4733-4743. Understanding the complex structure of multivariate extremes is a major challenge in various fields from portfolio monitoring and environmental risk management to insurance. In the framework of multivariate Extreme Value Theory, a common characterization of extremes' dependence structure is the angular measure. It is a suitable measure to work in extreme regions as it provides meaningful insights concerning the subregions where extremes tend to concentrate their mass. The present paper develops a novel optimization-based approach to assess the dependence structure of extremes. This support identification scheme rewrites as estimating clusters of features which best capture the support of extremes. The dimension reduction technique we provide is applied to statistical learning tasks such as feature clustering and anomaly detection. Numerical experiments provide strong empirical evidence of the relevance of our approach.
• Risks and security of internet and systems
  
  • Garcia‐alfaro Joaquin
  • Leneutre Jean
  • Cuppens Nora
  • Yaich Reda
  , 2021, 12528, pp.xi-378. This book constitutes the proceedings of the 15th International Conference on Risks and Security of Internet and Systems, CRiTIS 2020, which took place during November 4-6, 2020. The conference was originally planned to take place in Paris, France, but had to change to an online format due to the COVID-19 pandemic. The 16 full and 7 short papers included in this volume were carefully reviewed and selected from 44 submissions. In addition, the book contains one invited talk in full paper length. The papers were organized in topical sections named: vulnerabilities, attacks and intrusion detection; TLS, openness and security control; access control, risk assessment and security knowledge; risk analysis, neural networks and Web protection; infrastructure security and malware detection. (10.1007/978-3-030-68887-5)
  DOI : 10.1007/978-3-030-68887-5
• Optimal transport between determinantal point processes and application to fast simulation
  
  • Decreusefond Laurent
  • Moroz Guillaume
  Modern Stochastics: Theory and Applications, VTEX, 2021, 8 (2), pp.209--237. We analyze several optimal transportation problems between de-terminantal point processes. We show how to estimate some of the distances between distributions of DPP they induce. We then apply these results to evaluate the accuracy of a new and fast DPP simulation algorithm. We can now simulate in a reasonable amount of time more than ten thousands points. (10.15559/21-VMSTA180)
  DOI : 10.15559/21-VMSTA180
• Phoneme Level Lyrics Alignment and Text-Informed Singing Voice Separation
  
  • Schulze-Forster Kilian
  • Doire Clement S J
  • Richard Gael
  • Badeau Roland
  IEEE/ACM Transactions on Audio, Speech and Language Processing, Institute of Electrical and Electronics Engineers, 2021. The goal of singing voice separation is to recover the vocals signal from music mixtures. State-of-the-art performance is achieved by deep neural networks trained in a supervised fashion. Since training data are scarce and music signals are extremely diverse, it remains challenging to achieve high separation quality across various recording and mixing conditions as well as music styles. In this paper, we investigate to which extent the separation can be improved when lyrics transcripts are used as additional information. To this end, we propose a joint approach to phoneme level lyrics alignment and text-informed singing voice separation. It is based on DTW-attention, a new monotonic attention mechanism including a differentiable approximation of dynamic time warping. Experimental results show that the method can align phonemes with mixed singing voice with high precision given accurate transcripts. It also achieves competitive results on challenging word level alignment test sets using less training data than state-of-the-art methods. Sequential alignment and informed separation lead to improved separation quality according to objective measures. Text information helps preserving spectral phoneme properties in the separated voice signals. (10.1109/TASLP.2021.3091817)
  DOI : 10.1109/TASLP.2021.3091817
• The Vagueness of Vagueness in Noun Phrases
  
  • Paris Pierre-Henri
  • El Aoud Syrine
  • Suchanek Fabian
  , 2021. Natural language text has a great potential to feed knowledge bases. However, natural language is not always precise-and sometimes intentionally so. In this position paper, we study vagueness in noun phrases. We manually analyze the frequency of vague noun phrases in a Wikipedia corpus, and find that 1/4 of noun phrases exhibit some form of vagueness. We report on their nature and propose a categorization. We then conduct a literature review and present different definitions of vagueness, and different existing methods to deal with the detection and modeling of vagueness. We find that, despite its frequency, vagueness has not yet be addressed in its entirety.
• Les mégadonnées et l'essor de l'intelligence artificielle
  
  • Clémençon Stéphan
  Les Cahiers français : documents d'actualité, La Documentation Française, 2021 (419), pp.68.
• Computer-aided diagnosis tool for cervical cancer screening with weakly supervised localization and detection of abnormalities using adaptable and explainable classifier
  
  • Pirovano A.
  • Almeida Leandro G
  • Ladjal Saïd
  • Bloch Isabelle
  • Berlemont S.
  Medical Image Analysis, Elsevier, 2021, 73, pp.102167. While pap test is the most common diagnosis methods for cervical cancer, their results are highly dependent on the ability of the cytotechnicians to detect abnormal cells on the smears using brightfield microscopy. In this paper, we propose an explainable region classifier in whole slide images that could be used by cyto-pathologists to handle efficiently these big images (100,000x100,000 pixels). We create a dataset that simulates pap smears regions and uses a loss, we call classification under regression constraint, to train an efficient region classifier (about 66.8% accuracy on severity classification, 95.2% accuracy on normal/abnormal classification and 0.870 KAPPA score). We explain how we benefit from this loss to obtain a model focused on sensitivity and, then, we show that it can be used to perform weakly supervised localization (accuracy of 80.4%) of the cell that is mostly responsible for the malignancy of regions of whole slide images. We extend our method to perform a more general detection of abnormal cells (66.1% accuracy) and ensure that at least one abnormal cell will be detected if malignancy is present. Finally, we experiment our solution on a small real clinical slide dataset, highlighting the relevance of our proposed solution, adapting it to be as easily integrated in a pathology laboratory workflow as possible, and extending it to make a slide-level prediction. (10.1016/j.media.2021.102167)
  DOI : 10.1016/j.media.2021.102167
• Self-healing Networks via Self-organising Mobile Agents
  
  • Rodriguez Arles
  • Gomez Jonatan
  • Diaconescu Ada
  Journal of Autonomous Agents and Multi-agent Systems (JAAMAS), 2021.
• DNN-based mask estimation for distributed speech enhancement in spatially unconstrained microphone arrays
  
  • Furnon Nicolas
  • Serizel Romain
  • Essid Slim
  • Illina Irina
  IEEE/ACM Transactions on Audio, Speech and Language Processing, Institute of Electrical and Electronics Engineers, 2021, 29, pp.2310 - 2323. Deep neural network (DNN)-based speech enhancement algorithms in microphone arrays have now proven to be efficient solutions to speech understanding and speech recognition in noisy environments. However, in the context of ad-hoc microphone arrays, many challenges remain and raise the need for distributed processing. In this paper, we propose to extend a previously introduced distributed DNN-based time-frequency mask estimation scheme that can efficiently use spatial information in form of so-called compressed signals which are pre-filtered target estimations. We study the performance of this algorithm named Tango under realistic acoustic conditions and investigate practical aspects of its optimal application. We show that the nodes in the microphone array cooperate by taking profit of their spatial coverage in the room. We also propose to use the compressed signals not only to convey the target estimation but also the noise estimation in order to exploit the acoustic diversity recorded throughout the microphone array. (10.1109/TASLP.2021.3092838)
  DOI : 10.1109/TASLP.2021.3092838
• Automatic Feature Selection for Improved Interpretability on Whole Slide Imaging
  
  • Pirovano A.
  • Heuberger H.
  • Berlemont S.
  • Ladjal Saïd
  • Bloch Isabelle
  Machine Learning and Knowledge Extraction, MDPI, 2021, 3 (1), pp.243-262. Deep learning methods are widely used for medical applications to assist medical doctors in their daily routine. While performances reach expert's level, interpretability (highlighting how and what a trained model learned and why it makes a specific decision) is the next important challenge that deep learning methods need to answer to be fully integrated in the medical field. In this paper, we address the question of interpretability in the context of whole slide images (WSI) classification with the formalization of the design of WSI classification architectures and propose a piece-wise interpretability approach, relying on gradient-based methods, feature visualization and multiple instance learning context. After training two WSI classification architectures on Camelyon-16 WSI dataset, highlighting discriminative features learned, and validating our approach with pathologists, we propose a novel manner of computing interpretability slide-level heat-maps, based on the extracted features, that improves tile-level classification performances. We measure the improvement using the tile-level AUC that we called Localization AUC, and show an improvement of more than 0.2. We also validate our results with a RemOve And Retrain (ROAR) measure. Then, after studying the impact of the number of features used for heat-map computation, we propose a corrective approach, relying on activation colocalization of selected features, that improves the performances and the stability of our proposed method. (10.3390/make3010012)
  DOI : 10.3390/make3010012
• Power Allocation for Uplink Multiband Satellite Communications with Nonlinear Impairments
  
  • Louchart Arthur
  • Ciblat Philippe
  • Poulliat Charly
  IEEE Communications Letters, Institute of Electrical and Electronics Engineers, 2021, 25 (8), pp.2713-2717. In this letter, we develop some generic power allocation strategies in an uplink multiband satellite communications system when nonlinear impairments on the High-Power Amplifier onboard satellite occur. Based on the capacity closed-form expression related to receivers seeing nonlinear interference as a noise, we propose practical and scalable algorithms for three power allocation problems: i) sum-power minimization, ii) maximization of minimum per-user data rate, iii) sum-rate maximization. We show that the solutions mainly rely on Geometric Programming and/or Successive Convex Approximation approaches. The proposed solutions outperform naive approaches while enabling user scalability contrary to optimal brute-force grid search algorithms. (10.1109/LCOMM.2021.3087408)
  DOI : 10.1109/LCOMM.2021.3087408
• River: machine learning for streaming data in Python
  
  • Montiel Jacob
  • Halford Max
  • Mastelini Saulo Martiello
  • Bolmier Geoffrey
  • Sourty Raphaël
  • Vaysse Robin
  • Zouitine Adil
  • Gomes Heitor Murilo
  • Read Jesse
  • Abdessalem Talel
  • Bifet Albert
  Journal of Machine Learning Research, Microtome Publishing, 2021, 22, pp.1-8. River is a machine learning library for dynamic data streams and continual learning. It provides multiple state-of-the-art learning methods, data generators/transformers, performance metrics and evaluators for different stream learning problems. It is the result from the merger of two popular packages for stream learning in Python: Creme and scikitmultiflow. River introduces a revamped architecture based on the lessons learnt from the seminal packages. River’s ambition is to be the go-to library for doing machine learning on streaming data. Additionally, this open source package brings under the same umbrella a large community of practitioners and researchers. The source code is available at https://github.com/online-ml/river. (10.48550/arXiv.2012.04740)
  DOI : 10.48550/arXiv.2012.04740
• Distributed Learning assisted Fronthaul Compression for Multi-Antenna Uplink C-RAN
  
  • Askri Aymen
  • Zhang Chao
  • Rekaya-Ben Othman Ghaya
  IEEE Access, IEEE, 2021.
• Un robot capable de calculer sa responsabilité sera-t-il responsable de ses actes?
  
  • Dessalles Jean-Louis
  , 2021. Il peut être choquant d’imaginer que des notions comme la responsabilité, l’intention, le jugement ou la négligence puissent faire l’objet de calculs. Or la décision juridique n’est pas ineffable, puisqu’elle est censée être motivée après coup par référence à des principes. Peut-on traduire ces principes sous une forme utilisable par des machines ?
• Probabilistic semi-nonnegative matrix factorization: a Skellam-based framework
  
  • Fuentes Benoît
  • Richard Gael
  Computing Research Repository, ACM / ArXiv, 2021. We present a new probabilistic model to address semi-nonnegative matrix factorization (SNMF), called Skellam-SNMF. It is a hierarchical generative model consisting of prior components, Skellam-distributed hidden variables and observed data. Two inference algorithms are derived: Expectation-Maximization (EM) algorithm for maximum \emph{a posteriori} estimation and Variational Bayes EM (VBEM) for full Bayesian inference, including the estimation of parameters prior distribution. From this Skellam-based model, we also introduce a new divergence D between a real-valued target data x and two nonnegative parameters λ0 and λ1 such that D(x∣λ0,λ1)=0⇔x=λ0−λ1, which is a generalization of the Kullback-Leibler (KL) divergence. Finally, we conduct experimental studies on those new algorithms in order to understand their behavior and prove that they can outperform the classic SNMF approach on real data in a task of automatic clustering.
• BAYESIAN NODE CLASSIFICATION FOR NOISY GRAPHS
  
  • Hafidi Hakim
  • Ghogho Mounir
  • Ciblat Philippe
  • Swami Ananthram
  , 2021. Graph neural networks (GNN) have been recognized as powerful tools for learning representations in graph structured data. The key idea is to propagate and aggregate information along edges of the given graph. However, little work has been done to analyze the effect of noise on their performance. By conducting a number of simulations, we show that GNN are very sensitive to the graph noise. We propose a graphassisted Bayesian node classifier which takes into account the degree of impurity of the graph, and show that it consistently outperforms GNN based classifiers on benchmark datasets, particularly when the degree of impurity is moderate to high.
• Infinite-dimensional gradient-based descent for alpha-divergence minimisation
  
  • Daudel Kamélia
  • Douc Randal
  • Portier François
  Annals of Statistics, Institute of Mathematical Statistics, 2021, 49 (4), pp.2250 - 2270. This paper introduces the $(\alpha, \Gamma)$-descent, an iterative algorithm which operates on measures and performs $\alpha$-divergence minimisation in a Bayesian framework. This gradient-based procedure extends the commonly-used variational approximation by adding a prior on the variational parameters in the form of a measure. We prove that for a rich family of functions $\Gamma$, this algorithm leads at each step to a systematic decrease in the $\alpha$-divergence and derive convergence results. Our framework recovers the Entropic Mirror Descent algorithm and provides an alternative algorithm that we call the Power Descent. Moreover, in its stochastic formulation, the $(\alpha, \Gamma)$-descent allows to optimise the mixture weights of any given mixture model without any information on the underlying distribution of the variational parameters. This renders our method compatible with many choices of parameters updates and applicable to a wide range of Machine Learning tasks. We demonstrate empirically on both toy and real-world examples the benefit of using the Power descent and going beyond the Entropic Mirror Descent framework, which fails as the dimension grows.
• A Latent Transformer for Disentangled Face Editing in Images and Videos
  
  • Yao Xu
  • Newson Alasdair
  • Gousseau Yann
  • Hellier Pierre
  , 2021, pp.13789-13798.
• Screening Rules and its Complexity for Active Set Identification
  
  • Ndiaye Eugene
  • Fercoq Olivier
  • Salmon Joseph
  Journal of Convex Analysis, Heldermann, 2021, 28 (4), pp.1053--1072. Screening rules were recently introduced as a technique for explicitly identifying active structures such as sparsity, in optimization problem arising in machine learning. This has led to new methods of acceleration based on a substantial dimension reduction. We show that screening rules stem from a combination of natural properties of subdifferential sets and optimality conditions, and can hence be understood in a unified way. Under mild assumptions, we analyze the number of iterations needed to identify the optimal active set for any converging algorithm. We show that it only depends on its convergence rate. (10.48550/arXiv.2009.02709)
  DOI : 10.48550/arXiv.2009.02709
• Approximate Inference and Learning of State Space Models with Laplace Noise
  
  • Neri Julian
  • Depalle Philippe
  • Badeau Roland
  IEEE Transactions on Signal Processing, Institute of Electrical and Electronics Engineers, 2021, 69, pp.3176 - 3189. State space models have been extensively applied to model and control dynamical systems in disciplines including neuroscience, target tracking, and audio processing. A common modeling assumption is that both the state and data noise are Gaussian because it simplifies the estimation of the system's state and model parameters. However, in many real-world scenarios where the noise is heavy-tailed or includes outliers, this assumption does not hold, and the performance of the model degrades. In this aper, we present a new approximate inference algorithm for state space models with Laplace-distributed multivariate data that is robust to a wide range of non-Gaussian noise. Exact inference is combined with an expectation propagation algorithm, leading to filtering and smoothing that outperforms existing approximate inference methods for Laplace-distributed data, while retaining a fast speed similar to the Kalman filter. Further, we present a maximum posterior expectation-maximization (EM) algorithm that learns the parameters of the model in an unsupervised way, automatically avoids over-fitting the data, and provides better model estimation than existing methods for the Gaussian model. The quality of the inference and learning algorithms are exemplified through a diverse set of experiments and an application to non-linear tracking of audio frequency. (10.1109/tsp.2021.3075146)
  DOI : 10.1109/tsp.2021.3075146
• Resolution of a Routing and Wavelength Assignment Problem by Independent Sets in Conflict Graphs
  
  • Hudry Olivier
  , 2021.
• Méta-apprentissage : classification de messages en catégories émotionnelles inconnues en entraînement
  
  • Guibon Gaël
  • Labeau Matthieu
  • Flamein Hélène
  • Lefeuvre Luce
  • Clavel Chloé
  , 2021, pp.199-208. Dans cet article nous reproduisons un scénario d’apprentissage selon lequel les données cibles ne sont pas accessibles et seules des données connexes le sont. Nous utilisons une approche par méta-apprentissage afin de déterminer si les méta-informations apprises à partir de messages issus de médias sociaux, finement annotés en émotions, peuvent produire de bonnes performances une fois utilisées sur des messages issus de conversations, étiquetés en émotions avec une granularité différente. Nous mettons à profit l’apprentissage sur quelques exemples (few-shot learning) pour la mise en place de ce scénario. Cette approche se montre efficace pour capturer les méta-informations d’un jeu d’étiquettes émotionnelles pour prédire des étiquettes jusqu’alors inconnues au modèle. Bien que le fait de varier le type de données engendre une baisse de performance, notre approche par méta-apprentissage atteint des résultats décents comparés au référentiel d’apprentissage supervisé.
• Sum-capacity of Uplink Multiband Satellite Communications with Nonlinear Impairments
  
  • Louchart Arthur
  • Ciblat Philippe
  • Poulliat Charly
  , 2021. A compact and closed-form expression of capacity is derived for a uplink multiband satellite system in the presence of nonlinear interference. The nonlinear effect comes from the satellite high-power amplifier modeled by a Volterra series expansion. The derivations reveal that the nonlinear interference can provide a constructive power contribution that could be used to increase the transmission rate. Consequently, decoders designed by viewing this interference as only an additional noise are suboptimal. Numerical results confirm this claim and also shows that an appropriate power allocation amongst the subbands may be of interest.
• Dual Optimization for Kolmogorov Model Learning Using Enhanced Gradient Descent
  
  • Duan Qiyou
  • Ghauch Hadi
  • Kim Taejoon
  IEEE Transactions on Signal Processing, Institute of Electrical and Electronics Engineers, 2021. Data representation techniques have made a substantial contribution to advancing data processing and machine learning (ML). Improving predictive power was the focus of previous representation techniques, which unfortunately perform rather poorly on the interpretability in terms of extracting underlying insights of the data. Recently, Kolmogorov model (KM) was studied, which is an interpretable and predictable representation approach to learning the underlying probabilistic structure of a set of random variables. The existing KM learning algorithms using semi-definite relaxation with randomization (SDRwR) or discrete monotonic optimization (DMO) have, however, limited utility to big data applications because they do not scale well computationally. In this paper, we propose a computationally scalable KM learning algorithm, based on the regularized dual optimization combined with enhanced gradient descent (GD) method. To make our method more scalable to large-dimensional problems, we propose two acceleration schemes, namely, eigenvalue decomposition (EVD) elimination strategy and proximal EVD algorithm. When applied to big data applications, it is demonstrated that the proposed method can achieve compatible training/prediction performance with significantly reduced computational complexity; roughly two orders of magnitude improvement in terms of the time overhead, compared to the existing KM learning algorithms. Furthermore, it is shown that the accuracy of logical relation mining for interpretability by using the proposed KM learning algorithm exceeds 80%.
• The Role of Digital Technologies in Responding to the Grand Challenges of the Natural Environment: The Windermere Accord
  
  • Blair Gordon
  • Bassett Richard
  • Bastin Louis
  • Beevers L.
  • Borrajo Garcia Maribel
  • Brown Mike
  • Dance Sarah L
  • Diaconescu Ada
  • Edwards Elizabeth
  • Ferrario Maria Angela
  • Fraser Robert
  • Harriet Fraser
  Patterns, Cell Press Elsevier, 2021.