Publications

2023

• Hierarchical Design of Cyber-Physical Systems
  
  • Genius Daniela
  • Apvrille Ludovic
  , 2023, pp.117-124. Cyber-physical systems are based upon analog / digital hardware and software components. The splitting into functionalities and interaction between analog and digital parts should be considered as early as possible in the design phase, relying on formal verification or simulation. While many papers pretend to propose a modeling environment supporting them, only a few of them really address the different Models of Computation of these systems because they strongly differ. The paper explains how to generate a combined SystemC/SystemC AMS virtual prototype of the analog and mixed-signal parts of CPS directly from a SysML model featuring whole parts of CPS, thus reconciling near-circuit precision with more abstract analog and digital models. (10.5220/0011654400003402)
  DOI : 10.5220/0011654400003402
• It’s not Just What You Do but also When You Do It: Novel Perspectives for Informing Interactive Public Speaking Training
  
  • Biancardi Beatrice
  • Duan Yingjie
  • Chollet Mathieu
  • Clavel Chloé
  , 2023, pp.193-200. Most of the emerging public speaking training systems, while very promising, leverage temporal-aggregate features, which do not take into account the structure of the speech. In this paper, we take a different perspective, testing whether some well-known socio-cognitive theories, like first impressions or primacy and recency effect, apply in the distinct context of public speaking perception. We investigated the impact of the temporal location of speech slices (i.e., at the beginning, middle or end) on the perception of confidence and persuasiveness of speakers giving online movie reviews (the Persuasive Opinion Multimedia dataset). Results show that, when considering multi-modality, usually the middle part of speech is the most informative. Additional findings also suggest the interest to leverage local interpretability (by computing SHAP values) to provide feedback directly, both at a specific time (what speech part?) and for a specific behaviour modality or feature (what behaviour ?). This is a first step towards the design of more explainable and pedagogical interactive training systems. Such systems could be more efficient by focusing on improving the speaker’s most important behaviour during the most important moments of their performance, and by situating feedback at specific places within the total speech. (10.5220/0011680400003417)
  DOI : 10.5220/0011680400003417
• A Hierarchical Design Tool for SystemC AMS
  
  • Genius Daniela
  • Apvrille Ludovic
  , 2024, 2106, pp.3-28. The splitting into functionalities and interaction between analog and digital parts should be considered as early as possible in the design phase. We extended this methodology in former work, in order to take cyber-physical systems, based upon analog / digital hardware and software components, into account. The Models of Computation in these systems are not frequently tackled by many researchers due to their significant differences. However, as we show in this article, a SysML model can be directly used to generate a virtual prototype representing both the analog and mixed-signal parts of cyber-physical systems. For this, we rely on a hierarchical methodology, as certain analog components demand highly detailed designs. In this study, we introduce the capability to perform extensive semi-automatic design space exploration for these parts as well. A way to achieve this is to use parameters to automatically adapt the number of subsystems considered in the prototype. (10.1007/978-3-031-66339-0_1)
  DOI : 10.1007/978-3-031-66339-0_1
• Mutation of Formally Verified SysML Models
  
  • Apvrille Ludovic
  • Sultan Bastien
  • Hotescu Oana
  • de Saqui-Sannes Pierre
  • Coudert Sophie
  , 2023. Model checking of SysML models contributes to detect design errors and to check design decisions against user requirements. Yet, each time a model is modified, formal verification must be performed again, which makes model evolution costly and hampers the use of agile development methods. Based on former contributions on dependency graphs, the paper proposes to facilitate updates (also called mutations) on models: whenever a mutation is performed on a model, the algorithms introduced in this paper can determine which proofs remain valid and which ones must be performed again. The main idea to reduce the proof obligation is to identify new paths that need to be re-verified. Our algorithm reuses the results of previous proofs as much as possible in order to lower the complexity of the proof. The paper focuses on reachability proofs. A real-time communication architecture based on TSN (Time Sensitive Networking) illustrates the approach and performance results are presented.
• Macrolitter video counting on riverbanks using state space models and moving cameras
  
  • Chagneux Mathis
  • Le Corff Sylvain
  • Gloaguen Pierre
  • Ollion Charles
  • Lepâtre Océane
  • Bruge Antoine
  Computo, Société Française de Statistique, 2023. Litter is a known cause of degradation in marine environments and most of it travels in rivers before reaching the oceans. In this paper, we present a novel algorithm to assist waste monitoring along watercourses. While several attempts have been made to quantify litter using neural object detection in photographs of floating items, we tackle the more challenging task of counting directly in videos using boat-embedded cameras. We rely on multi-object tracking (MOT) but focus on the key pitfalls of false and redundant counts which arise in typical scenarios of poor detection performance. Our system only requires supervision at the image level and performs Bayesian filtering via a state space model based on optical flow. We present a new open image dataset gathered through a crowdsourced campaign and used to train a center-based anchor-free object detector. Realistic video footage assembled by water monitoring experts is annotated and provided for evaluation. Improvements in count quality are demonstrated against systems built from state-of-the-art multi-object trackers sharing the same detection capabilities. A precise error decomposition allows clear analysis and highlights the remaining challenges. (10.57750/845m-f805)
  DOI : 10.57750/845m-f805
• On Several Mathematical and Data-Driven Models for Image and Video Editing, Synthesis and Analysis
  
  • Newson Alasdair
  , 2023. In this document, I present several mathematical and data-based models for image editing, synthesis and analysis. Firstly, I look at low-rank models used for background/foreground separation in videos. This model separates a video into the sum of a low-rank background and a sparse foreground, via an optimisation problem. I propose two algorithms using this model: firstly, a multi-temporal foreground separation algorithm and secondly a video segmentation method to identify regions where this model applies well. I then present a model to synthesis silver-halide film grain texture in digital images, based on stochastic geometry. This model is physically realistic and is able to synthesise grain at any output resolution, which gives high quality results. I also present an approximation to this model using Gaussian textures, with the goal of speeding up the synthesis algorithm. Finally, I discuss four works concerning deep generative models. The first is a mathematical analysis of how autoencoders, a type of deep learning model, can encode and decode simple geometric shapes. Secondly, I present a feed-forward neural network which edits the age of facial images. I then present an algorithm which uses a pre-trained deep generative model to edit general attributes of facial images. The method learns to navigate in the latent space of the generative model to achieve editing of the desired attribute. Finally, I present a ``PCA-Autoencoder'', which imitates the behaviour of the Principal Component Analysis (PCA) method, via a progressive increase of the latent space size, and a carefully chosen correlation loss function. I show how this can be used to carry out unsupervised editing of attributes in images.
• Local sampling of the SU(1,1) Wigner function
  
  • Fabre Nicolas
  • Klimov Andrei B
  • Leuchs Gerd
  • Sánchez-Soto Luis L
  AVS Quantum Science, AIP Publishing / AVS, 2023, 5 (1), pp.014404. Despite its indisputable merits, the Wigner phase-space formulation has not been widely explored for systems with SU(1,1) symmetry, as a simple operational definition of the Wigner function has proved elusive in this case. We capitalize on unique properties of the parity operator, to derive in a consistent way a bona fide SU(1,1) Wigner function that faithfully parallels the structure of its continuous-variable counterpart. We propose an optical scheme, involving a squeezer and photon-number-resolving detectors, that allows for direct point-by-point sampling of that Wigner function. This provides an adequate framework to represent SU(1,1) states satisfactorily. (10.1116/5.0134784)
  DOI : 10.1116/5.0134784
• The role of Mrs. Gerber's lemma for evaluating the information leakage of masked computations
  
  • Rioul Olivier
  , 2023. In the context of secret sharing computation in some finite Abelian group, given noisy observations of each share, how can one measure the information leakage of the secret? We show that in various instances of this problem, it boils down to establishing some variation of a ”Mrs. Gerber’s lemma”. That is, find a lower bound on some randomness measure of a sum of discrete random variables in the Abelian group in terms of the product of individual randomnesses of each discrete random variable. It is an open problem to generalize these approaches in a suitable framework.
• Teleportation of continuous variables states
  
  • Fabre N
  , 2023, pp.4.
• A Low-Profile, Triple-Band, and Wideband Antenna Using Dual-Band AMC
  
  • Gonçalves Licursi de Mello Rafael
  • Lepage Anne Claire
  • Begaud Xavier
  Sensors, MDPI, 2023, 23 (4), pp.1920 (1-19). When a wideband antenna is backed by an artificial magnetic conductor (AMC) reflector, the bandwidth is reduced. With the optimization of the shape of the AMC it is possible to exhibit multiband behavior, but the problem becomes complex if the bands are also intended to be wide. In this study, a methodology that exploits both the expected in-band and out-of-band behaviors of a dual-band AMC was used to design a low-profile, triple-band, and wideband directive antenna. The methodology was validated with a prototype suitable for the European standards of 4G/5G and Wi-Fi 2.4/5/6E, operating within the following bands: 2.4–2.7 GHz, 3.4–3.8 GHz, and 5.17–6.45 GHz. The measured results showed respective peak values of 8.0, 9.1, and 10.5 dBi for the broadside realized gain, front-to-back ratios larger than 19 dB, cross-polarized levels lower than -18 dB, and stable half-power beamwidths within each band. Furthermore, 3 dB gain bandwidths of 34.4%, 19.7%, and 31.0% were also measured. (10.3390/s23041920)
  DOI : 10.3390/s23041920
• Automatically Verifying Expressive Epistemic Properties of Programs
  
  • Belardinelli Francesco
  • Boureanu Ioana
  • Malvone Vadim
  • Rajaona Fortunat
  , 2023, 37 (5), pp.6245-6252. We propose a new approach to the verification of epistemic properties of programmes. First, we introduce the new ``program-epistemic'' logic L_PK, which is strictly richer and more general than similar formalisms appearing in the literature. To solve the verification problem in an efficient way, we introduce a translation from our language L_PK into first-order logic. Then, we show and prove correct a reduction from the model checking problem for program-epistemic formulas to the satisfiability of their first-order translation. Both our logic and our translation can handle richer specification w.r.t. the state of the art, allowing us to express the knowledge of agents about facts pertaining to programs (i.e., agents' knowledge before a program is executed as well as after is has been executed). Furthermore, we implement our translation in Haskell in a general way (i.e., independently of the programs in the logical statements), and we use existing SMT-solvers to check satisfaction of L_PK formulas on a benchmark example in the AI/agency field. (10.1609/AAAI.V37I5.25769)
  DOI : 10.1609/AAAI.V37I5.25769
• An Adaptive Layer to Leverage Both Domain and Task Specific Information from Scarce Data
  
  • Guibon Gaël
  • Labeau Matthieu
  • Lefeuvre Luce
  • Clavel Chloé
  , 2023, 37 (6), pp.7757-7765. Many companies make use of customer service chats to help the customer and try to solve their problem. However, customer service data is confidential and as such, cannot easily be shared in the research community. This also implies that these data are rarely labeled, making it difficult to take advantage of it with machine learning methods. In this paper we present the first work on a customer’s problem status prediction and identification of problematic conversations. Given very small subsets of labeled textual conversations and unlabeled ones, we propose a semi-supervised framework dedicated to customer service data leveraging speaker role information to adapt the model to the domain and the task using a two-step process. Our framework, Task-Adaptive Fine-tuning, goes from predicting customer satisfaction to identifying the status of the customer’s problem, with the latter being the main objective of the multi-task setting. It outperforms recent inductive semi-supervised approaches on this novel task while only considering a relatively low number of parameters to train on during the final target task. We believe it can not only serve models dedicated to customer service but also to any other application making use of confidential conversational data where labeled sets are rare. Source code is available at https://github.com/gguibon/taft (10.1609/aaai.v37i6.25940)
  DOI : 10.1609/aaai.v37i6.25940
• Graph-Assisted Bayesian Node Classifiers
  
  • Hafidi Hakim
  • Ciblat Philippe
  • Ghogho Mounir
  • Swami Ananthram
  IEEE Access, IEEE, 2023, 11, pp.23989-24002. Many datasets can be represented by attributed graphs on which classification methods may be of interest. The problem of node classification has attracted the attention of scholars due to its wide range of applications. The problem consists of predicting nodes' labels based on their intrinsic features, features of their neighboring nodes and the graph structure. Graph Neural Networks (GNN) have been widely used to tackle this task. Thanks to the graph structure and the node features, they are able to propagate information over the graph and aggregate it to improve the classification performance. Their performance is however sensitive to the graph topology, especially its degree of impurity, a measure of the proportion of connected nodes belonging to different classes. Here, we propose a new Graph-Assisted Bayesian (GAB) classifier, which is designed for the problem of node classification. By using the Bayesian theorem, GAB takes into consideration the degree of impurity of the graph when classifying the nodes. We show that the proposed classifier is less sensitive to graph impurity, and less complex than GNN-based classifiers. (10.1109/ACCESS.2023.3242866)
  DOI : 10.1109/ACCESS.2023.3242866
• Robust machine learning for Graphs/Networks
  
  • Hafidi Hakim
  , 2023. This thesis addresses advancements in graph representation learning, focusing on the challengesand opportunities presented by Graph Neural Networks (GNNs). It highlights the significanceof graphs in representing complex systems and the necessity of learning node embeddings that capture both node features and graph structure. The study identifies key issues in GNNs, such as their dependence on high-quality labeled data, inconsistent performanceacross various datasets, and susceptibility to adversarial attacks.To tackle these challenges, the thesis introduces several innovative approaches. Firstly, it employs contrastive learning for node representation, enabling self-supervised learning that reduces reliance on labeled data. Secondly, a Bayesian-based classifier isproposed for node classification, which considers the graph’s structure to enhance accuracy. Lastly, the thesis addresses the vulnerability of GNNs to adversarialattacks by assessing the robustness of the proposed classifier and introducing effective defense mechanisms.These contributions aim to improve both the performance and resilience of GNNs in graph representation learning. (10.70675/cae1cb8bz81a2z4215z8817zfff07c056862)
  DOI : 10.70675/cae1cb8bz81a2z4215z8817zfff07c056862
• Devices and methods for low power adaptive channel sensing
  
  • Tchamkerten Aslan
  • Jabbour Chadi
  • Mukherjee Manuj
  , 2023.
• E. coli et compagnie
  
  • Zayana Karim
  • Boyer Ivan
  • Lartigue Christophe
  Bulletin de l'Union des Professeurs de Spéciales, Union des Professeurs de Spéciales, 2023, Hiver 2022-2023. Based on a very simple probabilistic model of the evolution of a population of bacteria, the article crosses the biology and mathematics programs from high school to bachelor's degree and, in doing so, opens up to some health and of society.
• Modelling of daily radiofrequency electromagnetic field dose for a prospective adolescent cohort
  
  • Eeftens Marloes
  • Shen Chen
  • Sönksen Jana
  • Schmutz Claudia
  • van Wel Luuk
  • Liorni Ilaria
  • Vermeulen Roel
  • Cardis Elisabeth
  • Wiart Joe
  • Toledano Mireille
  • Röösli Martin
  Environment International, Elsevier, 2023, 172, pp.107737. (10.1016/j.envint.2023.107737)
  DOI : 10.1016/j.envint.2023.107737
• Full parameter extraction of a temperature-insensitive quantum well DFB laser using an optical injection technique
  
  • Ding Shihao
  • Doggett Nate
  • Herrera Daniel
  • Huang Heming
  • Kovanis Vassilios
  • Lester Luke
  • Grillot Frédéric
  , 2023, pp.23. Distributed feedback lasers are key ingredients of high-speed, high-capacity integrated photonic chips. In this work, we extract the linewidth enhancement factor above threshold by measuring the transitional points in the optical-injection stability map from a quantum well distributed feedback laser with a temperature-controlled mismatch between the lasing and optical gain peaks. This unique measurement technique allows the simultaneous extraction of important parameters influencing the linewidth, particularly the photon lifetime. When the current is higher than twice threshold and 50 ℃, the linewidth enhancement factor is smaller than that at 10 ℃. This effect is attributed to the increasing differential gain at the lasing peak position, which is a result of the larger optical mismatch. We also measured the spectral linewidth at different temperatures, which then yields the spontaneous emission factor, nsp. Due to the low linewidth enhancement factor at high temperatures, a large photon lifetime, and a modest increase in nsp, the linewidth does not drastically increase with pump current and stays below 100 kHz at 50 ℃. Overall, the stability of the linewidth enhancement factor combined with the large optical mismatch brings a relative temperature insensitivity, which is of paramount importance for applications requiring high-temperature operation and improved coherent light. (10.1117/12.2650460)
  DOI : 10.1117/12.2650460
• Accurate prediction of performance for a submarine open cable
  
  • Girard-Jollet Joana
  • Antona Jean-Christophe
  • Carbo Meseguer Alexis
  • Lonardi Matteo
  • Ramantanis Petros
  • Rekaya Ghaya
  • Le Meur Gaëlle
  • Leplingard Florence
  • Lenne Edmond
  • Olsson Samuel
  , 2023, pp.22. (10.1117/12.2648763)
  DOI : 10.1117/12.2648763
• Dynamic and nonlinear properties of mid-infrared interband quantum cascade lasers
  
  • Grillot Frédéric
  • Spitz Olivier
  • Zhao Shiyuan
  • Didier Pierre
  , 2023, pp.15. Interband cascade lasers (ICLs) constitute a new class of semiconductor lasers allowing lasing emission in the 3– 7 μm wavelength region. Their structure presents similarities and differences with respect to both standard bipolar semiconductor lasers and quantum cascade lasers (QCLs). In contrast to QCLs, the stimulated emission of ICLs relies on the interband transition of type-II quantum wells while the carrier-to-photon lifetime ratio is similar to conventional bipolar lasers. ICLs can be classified into class-B laser systems like common quantum well lasers, and they exhibit a multi-GHz relaxation oscillation frequency that is related to the maximum modulation/chaos bandwidth achievable by these lasers. Moreover, ICLs take advantage of a cascading mechanism over repeated active regions, which allows us to boost the quantum efficiency and, thus, the emitted optical power. On top of that, the power consumption of ICLs is one or two orders of magnitude lower than their QCL counterparts whereas high-power of few hundreds of milliWatts can be achieved. Here, we report some recent results on the dynamic and nonlinear properties of ICLs. In particular, we demonstrate the generation of fully-developed chaos under external optical feedback. We show that ICLs exhibit some peculiar intensity noise features with a clear relaxation oscillation frequency. Together, these properties are of paramount importance for developing long-reach secure free-space communication, random bit generator, and remote chaotic LiDAR systems. Lastly, we also predict that ICLs are preferable devices for amplitude-noise squeezing because large amplitude noise reduction is attainable through inherent high quantum efficiency and short photon and electron lifetimes. (10.1117/12.2651426)
  DOI : 10.1117/12.2651426
• Recent advances in high-speed data communications using mid-infrared quantum cascade lasers
  
  • Grillot Frédéric
  • Spitz Olivier
  • Dely Hamza
  • Didier Pierre
  • Bonazzi Thomas
  • Awwad Elie
  • Vasanelli Angela
  • Sirtori Carlo
  , 2023.
• DNA Code from Cyclic and Skew Cyclic Codes over F4[v]/⟨v3⟩
  
  • Prakash Om
  • Singh Ashutosh
  • Verma Ram Krishna
  • Solé Patrick
  • Cheng Wei
  Entropy, MDPI, 2023, 25 (2), pp.239. The main motivation of this work is to study and obtain some reversible and DNA codes of length n with better parameters. Here, we first investigate the structure of cyclic and skew cyclic codes over the chain ring R:=F4[v]/⟨v3⟩. We show an association between the codons and the elements of R using a Gray map. Under this Gray map, we study reversible and DNA codes of length n. Finally, several new DNA codes are obtained that have improved parameters than previously known codes. We also determine the Hamming and the Edit distances of these codes. (10.3390/e25020239)
  DOI : 10.3390/e25020239
• Optimisation of the pre-compensation phase for GEO-feeder optical uplinks
  
  • Lognoné Perrine
  • Conan Jean-Marc
  • Rekaya Ghaya
  • Paillier Laurie
  • Védrenne Nicolas
  , 2023, pp.42. We propose a new MMSE method relying on phase and log-amplitude on-axis measurements and statistical priors to estimate the pre-compensation phase at point-ahead angle of a ground to geostationnary satellite telecom link suffering from anisoplanatism. This method shows to reduce the tip and tilt residual phase variance down to 49% and therefore brings a gain on the link margin up to 15 dB. It also shows to improve the fade statistics, reducing the number and mean duration of fades. (10.1117/12.2648898)
  DOI : 10.1117/12.2648898
• Contributions to the Design of Safe Complex Systems
  
  • Ameur Boulifa Rabéa
  , 2023. My research work is part of a framework that aims to develop formal approaches to help design complex systems with a good level of safety and security. More precisely, my work consists in proposing a theoretical and practical framework allowing to ease the modelling and verification of complex distributed and embedded systems. It is well known that the complexity of systems, especially concurrent systems, reliability issues and time-to-market constraints are examples of some of the current challenges that push existing system de- sign methodologies to their limits. These systems are characterised by a large number of interacting heterogeneous entities, described through different models (hardware and software), various programming languages, and use various means of communication. Moreover, these systems are often in perpetual evolution, the entities are designed with the objective of evolving and are intended to be extensible in order to cope with unplanned or unanticipated functionalities. Their analysis requires the combination of several points of view (different levels of specification and abstraction, various formalisms, etc). Faced with the complexity of models through their size, and the variety of interaction mechanisms, the challenge is the rigorous integration of these different aspects of a system within an unified and rigorous approach. In other words, the challenge is to define a procedure for integrating several perspectives within a single model. The models used so far for the modelling and behavioural verification of systems have revealed their limitations. For instance, most formalisms are not able to ”represent” the variability or parametric aspect of a multi-entities system. The ”extensibility” (and infinite) aspect is no less difficult to model. To address these issues, we propose a theoretical framework, that can help to model and verify complex (and concurrent) systems. This framework offers an approach supporting incremental design via symbolic, compositional, hierarchical modelling, and verification.
• Limitations of variational quantum algorithms: a quantum optimal transport approach
  
  • de Palma Giacomo
  • Marvian Milad
  • Rouzé Cambyse
  • Stilck França Daniel
  PRX Quantum, APS Physics, 2023, 4, pp.010309. The impressive progress in quantum hardware of the last years has raised the interest of the quantum computing community in harvesting the computational power of such devices. However, in the absence of error correction, these devices can only reliably implement very shallow circuits or comparatively deeper circuits at the expense of a nontrivial density of errors. In this work, we obtain extremely tight limitation bounds for standard NISQ proposals in both the noisy and noiseless regimes, with or without error-mitigation tools. The bounds limit the performance of both circuit model algorithms, such as QAOA, and also continuous-time algorithms, such as quantum annealing. In the noisy regime with local depolarizing noise $p$, we prove that at depths $L={\cal O}(p^{-1})$ it is exponentially unlikely that the outcome of a noisy quantum circuit outperforms efficient classical algorithms for combinatorial optimization problems like Max-Cut. Although previous results already showed that classical algorithms outperform noisy quantum circuits at constant depth, these results only held for the expectation value of the output. Our results are based on newly developed quantum entropic and concentration inequalities, which constitute a homogeneous toolkit of theoretical methods from the quantum theory of optimal mass transport whose potential usefulness goes beyond the study of variational quantum algorithms. (10.1103/PRXQuantum.4.010309)
  DOI : 10.1103/PRXQuantum.4.010309