Publications

2023

• Dynamical correlations in simple disorder and complex disorder liquids
  
  • Lovrinčević Bernarda
  • Požar Martina
  • Jukić Ivo
  • Perera David
  • Perera Aurélien
  Journal of Molecular Liquids, Elsevier, 2023, 393, pp.123421. Liquids in equilibrium exhibit two types of disorder, simple and com- plex. Typical simple disorder liquid are liquid nitrogen, or weakly polar liquids. Complex liquids concern those who can form long lived local as- semblies, and cover a large range from water to some soft matter and bio- logical liquids. The existence of such structures leaves characteric features upon the atom-atom correlation functions, concerning both atoms which directly participate to these structure and those who do not. The ques- tion we ask here is: do these features have also characteristic dynamical aspects, which could be tracked through dynamical correlation functions? Herein, we compare the van Hove function, intermediate scattering func- tion and the dynamical structure factor, for both types of liquids, using force eld models and computer simulations. The calculations reveal the paradoxical fact that neighbouring atom correlations for simple disorder liquids relax slower than that for complex disorder liquids, while prepeak features typical of complex disorder liquids relax even slower. This is an indication of the existence of fast kinetic self-assembly processes in com- plex disorder liquids, while the lifetime of such assemblies itself is quite slow. This is further conrmed by the existence of a very low-k dynamical pre-peak uncovered in the case of water and ethanol. (10.1016/j.molliq.2023.123421)
  DOI : 10.1016/j.molliq.2023.123421
• Mining bias-target Alignment from Voronoi Cells
  
  • Nahon Rémi
  • Nguyen Van-Tam
  • Tartaglione Enzo
  , 2023. Despite significant research efforts, deep neural networks are still vulnerable to biases: this raises concerns about their fairness and limits their generalization. In this paper, we propose a bias-agnostic approach to mitigate the impact of bias in deep neural networks. Unlike traditional debiasing approaches, we rely on a metric to quantify ``bias alignment/misalignment'' on target classes, and use this information to discourage the propagation of bias-target alignment information through the network. We conduct experiments on several commonly used datasets for debiasing and compare our method to supervised and bias-specific approaches. Our results indicate that the proposed method achieves comparable performance to state-of-the-art supervised approaches, although it is bias-agnostic, even in presence of multiple biases in the same sample.
• Physical-layer-aware network simulator for future optical functionalities
  
  • Garbhapu Venkata Virajit
  • Ware Cédric
  • Lourdiane Mounia
  , 2023. We developed a new optical network (ON) simulator that aims to take into account the physical layer impairments and eventually handle future optical functionalities (OFs) in a generic way. The tool is designed to accurately evaluate the Quality of Transmission (QoT), by modelling precisely the physical degradations affecting the transmitted signals. The QoT encompasses the physical impairments and is used in the routing and wavelength assignment (RWA) decision making at the network level. Preliminary results show the impact of an accurate evaluation of the physical impairments on the traffic blocking probability compared to results obtained without the QoT input. In addition, we give a practical example on how our simulator is able to integrate wavelength conversion, a well-known OF, and evaluate its impact on the network performance. This opens the perspective to add new OFs and propose a wider range of use cases.
• SCoTTi: Save Computation at Training Time with an adaptive framework
  
  • Li Ziyu
  • Tartaglione Enzo
  • Nguyen Van-Tam
  , 2023, pp.1435-1444. On-device training is an emerging approach in machine learning where models are trained on edge devices, aiming to enhance privacy protection and real-time performance. However, edge devices typically possess restricted computational power and resources, making it challenging to perform computationally intensive model training tasks. Consequently, reducing resource consumption during training has become a pressing concern in this field. To this end, we propose SCoTTi (Save Computation at Training Time), an adaptive framework that addresses the aforementioned challenge. It leverages an optimizable threshold parameter to effectively reduce the number of neuron updates during training which corresponds to a decrease in memory and computation footprint. Our proposed approach demonstrates superior performance compared to the state-of-the-art methods regarding computational resource savings on various commonly employed benchmarks and popular architectures, including ResNets, MobileNet, and Swin-T. (10.1109/ICCVW60793.2023.00156)
  DOI : 10.1109/ICCVW60793.2023.00156
• Few Labels are Enough! Semi-Supervised Graph Learning for Social Interaction
  
  • Corbellini Nicola
  • Giraldo Jhony H.
  • Varni Giovanna
  • Volpe Gualtiero
  , 2023, pp.3060-3068. Endowing machines with social intelligence is a fundamental goal of artificial social intelligence. Dealing with human-centered phenomena requires, however, a considerable amount of manually annotated data, making data annotation a costly and challenging task that hinders the training of supervised learning algorithms. In this study, we apply an approach grounded on Graph Convolutional Network (GCN) to alleviate the annotation burden. As a test bed, we select emergent states analysis with specific reference to the team potency. At first, we build the POTENCY dataset by fusing three datasets on social interaction. Next, we compute a set of multimodal features characterizing the social behavior of the team members and the team as one. Finally, we feed the POTENCY dataset to a semi-supervised GCN, trained on a binary node classification task, with variable amounts of labels. We show that GCN can assign team potency labels to an unlabeled team in the dataset by using only a few labeled examples (i.e., 10% of data), with performances comparable to or higher than those of two baseline algorithms carrying out the same task in a fully supervised way.
• Conférence Nationale d’Intelligence Artificielle Année 2023
  
  • Adam Emmanuel
  • Guyet Thomas
  • Le Blanc Benoit
  • Longin Dominique
  • Bouraoui Zied
  • Bringay Sandra
  • Gaudel Romaric
  • Laclau Charlotte
  • Launois Christelle
  • Morge Maxime
  • Roussey Catherine
  • Schwarzentruber François
  • Trojahn Cassia
  • Vareilles Élise
  • Wilczynski Anaelle
  , 2023. Cet ouvrage présente les activités de l'AFIA du 1er août 2021 au 31 juillet 2023, ainsi qu'une sélection d'article issus de la Plate-Forme Intelligence Artificielle de l'année (PFIA 2023).
• The Unreasonable Effectiveness of Large Language-Vision Models for Source-free Video Domain Adaptation
  
  • Zara Giacomo
  • Conti Alessandro
  • Roy Subhankar
  • Lathuilière Stéphane
  • Rota Paolo
  • Ricci Elisa
  , 2023. Source-Free Video Unsupervised Domain Adaptation (SFVUDA) task consists in adapting an action recognition model, trained on a labelled source dataset, to an unlabelled target dataset, without accessing the actual source data. The previous approaches have attempted to address SFVUDA by leveraging self-supervision (e.g., enforcing temporal consistency) derived from the target data itself. In this work, we take an orthogonal approach by exploiting "web-supervision" from Large Language-Vision Models (LLVMs), driven by the rationale that LLVMs contain a rich world prior surprisingly robust to domain-shift. We showcase the unreasonable effectiveness of integrating LLVMs for SFVUDA by devising an intuitive and parameter-efficient method, which we name Domain Adaptation with Large Language-Vision models (DALL-V), that distills the world prior and complementary source model information into a student network tailored for the target. Despite the simplicity, DALL-V achieves significant improvement over state-of-the-art SFVUDA methods.
• Shannon Strikes Again! Entropy-Based Pruning in Deep Neural Networks for Transfer Learning Under Extreme Memory and Computation Budgets
  
  • Spadaro Gabriele
  • Renzulli Riccardo
  • Bragagnolo Andrea
  • Giraldo Jhony H.
  • Fiandrotti Attilio
  • Grangetto Marco
  • Tartaglione Enzo
  , 2023, pp.1518-1522. Deep neural networks have become the de-facto standard across various computer science domains. Nonetheless, effectively training these deep networks remains challenging and resource-intensive. This paper investigates the efficacy of pruned deep learning models in transfer learning scenarios under extremely low memory budgets, tailored for TinyML models. Our study reveals that the source task's model with the highest activation entropy outperforms others in the target task. Motivated by this, we propose an entropy-based Efficient Neural Transfer with Reduced Overhead via PrunIng (ENTROPI) algorithm. Through comprehensive experiments on diverse models (ResNet18 and MobileNet-v3) and target datasets (CIFAR-100, VLCS, and PACS), we substantiate the superior generalization achieved by transfer learning from the entropy-pruned model. Quantitative measures for entropy provide valuable insights into the reasons behind the observed performance improvements. The results underscore ENTROPI's potential as an efficient solution for enhancing generalization in data-limited transfer learning tasks.
• Can Unstructured Pruning Reduce the Depth in Deep Neural Networks?
  
  • Liao Zhu
  • Quétu Victor
  • Nguyen Van-Tam
  • Tartaglione Enzo
  , 2023. Pruning is a widely used technique for reducing the size of deep neural networks while maintaining their performance. However, such a technique, despite being able to massively compress deep models, is hardly able to remove entire layers from a model (even when structured): is this an addressable task? In this study, we introduce EGP, an innovative Entropy Guided Pruning algorithm aimed at reducing the size of deep neural networks while preserving their performance. The key focus of EGP is to prioritize pruning connections in layers with low entropy, ultimately leading to their complete removal. Through extensive experiments conducted on popular models like ResNet-18 and Swin-T, our findings demonstrate that EGP effectively compresses deep neural networks while maintaining competitive performance levels. Our results not only shed light on the underlying mechanism behind the advantages of unstructured pruning, but also pave the way for further investigations into the intricate relationship between entropy, pruning techniques, and deep learning performance. The EGP algorithm and its insights hold great promise for advancing the field of network compression and optimization. The source code for EGP is released open-source.
• The Robust Semantic Segmentation UNCV2023 Challenge Results
  
  • Yu Xuanlong
  • Sanmiguel Juan C
  • Zhang Xiaowen
  • Escudero-Viñolo Marcos
  • Peng Rui
  • Tian Hanlin
  • Wang Xinyi
  • Matsui Kenta
  • Zhao Jiaxuan
  • Wang Tianhao
  • Zhang Junpei
  • Adan Fahmy
  • Wang Zitao
  • Gao Zhitong
  • He Xuming
  • Yang Yuting
  • Bouniot Quentin
  • Liu Fang
  • Moghaddam Hossein
  • Zuo Yi
  • Nandan Rai Shyam
  • Zhang Kexin
  • Cermelli Fabio
  • Alcover-Couso Roberto
  • Masone Carlo
  • Jiao Licheng
  • Pilzer Andrea
  • Ricci Elisa
  • Bursuc Andrei
  • Solin Arno
  • Trapp Martin
  • Li Rui
  • Yao Angela
  • Chen Wenlong
  • Simpson Ivor
  • Campbell Neill D. F.
  • Franchi Gianni
  , 2023, pp.4620-4630. This paper outlines the winning solutions employed in addressing the MUAD uncertainty quantification challenge held at ICCV 2023. The challenge was centered around semantic segmentation in urban environments, with a particular focus on natural adversarial scenarios. The report presents the results of 19 submitted entries, with numerous techniques drawing inspiration from cutting-edge uncertainty quantification methodologies presented at prominent conferences in the fields of computer vision and machine learning and journals over the past few years. Within this document, the challenge is introduced, shedding light on its purpose and objectives, which primarily revolved around enhancing the robustness of semantic segmentation in urban scenes under varying natural adversarial conditions. The report then delves into the top-performing solutions. Moreover, the document aims to provide a comprehensive overview of the diverse solutions deployed by all participants. By doing so, it seeks to offer readers a deeper insight into the array of strategies that can be leveraged to effectively handle the inherent uncertainties associated with autonomous driving and semantic segmentation, especailly within urban environments. (10.1109/iccvw60793.2023.00496)
  DOI : 10.1109/iccvw60793.2023.00496
• Morpho-Logic from a Topos Perspective – Application to Symbolic AI
  
  • Aiguier Marc
  • Bloch Isabelle
  • Nibouche Salim
  • Pino Pérez Ramón
  International Journal of Approximate Reasoning, Elsevier, 2023, 161, pp.109011. Modal logics have proved useful for many reasoning tasks in symbolic artificial intelligence (AI), such as belief revision, spatial reasoning, among others. On the other hand, mathematical morphology (MM) is a theory for non-linear analysis of structures, that was widely developed and applied in image analysis. Its mathematical bases rely on algebra, complete lattices, topology. Strong links have been established between MM and mathematical logics, mostly modal logics. In this paper, we propose to further develop and generalize this link between mathematical morphology and modal logic from a topos perspective, i.e. categorial structures generalizing space, and connecting logics, sets and topology. Furthermore, we rely on the internal language and logic of a topos. We define structuring elements, dilations and erosions as morphisms. Then we introduce the notion of structuring neighborhoods, and show that the dilations and erosions based on them lead to a constructive modal logic, for which a sound and complete proof system is proposed. We then show that the modal logic thus defined (called morpho-logic here), is well adapted to define concrete and efficient operators for revision, merging, and abduction of new knowledge, or even spatial reasoning. (10.1016/J.IJAR.2023.109011)
  DOI : 10.1016/J.IJAR.2023.109011
• All of Monty
  
  • Zayana Karim
  • Boyer Ivan
  Quadrature, EDP Sciences, 2023. Tout a déjà été dit sur le problème de Monty Hall, au point d’attirer l’attention du programme de mathématiques adossé à l’enseignement scientifique de première. L’objet de ce texte n’est donc pas d’en rajouter, mais plutôt d'y faire le tri tout en clarifiant quelques scénarios pédagogiques parus dans la littérature – foisonnante – déjà consacrée au sujet
• The Self-Anti-Censorship Nature of Encryption: On the Prevalence of Anamorphic Cryptography
  
  • Kutylowski Miroslaw
  • Persiano Giuseppe
  • Phan Duong Hieu
  • Yung Moti
  • Zawada Marcin
  Proceedings on Privacy Enhancing Technologies, Privacy Enhancing Technologies Symposium, 2023, 2023 (4), pp.170-183. As part of the responses to the ongoing crypto wars, the notion of Anamorphic Encryption was put forth. The notion allows private communication in spite of a dictator who is engaged in an extreme form of surveillance and or censorship, where it asks for all private keys and knows and may even dictate all messages. The original work pointed out efficient ways to use two known schemes in the anamorphic mode, bypassing the draconian censorship and hiding information from the all-powerful dictator. A question left open was whether these examples are outlier results or whether anamorphic mode is pervasive in existing systems. Here we answer the above question: we develop new techniques, expand the notion, and show that the notion of Anamorphic Cryptography is, in fact, very much prevalent. We first refine the notion of Anamorphic Encryption with respect to the nature of covert communication. Specifically, we distinguish Single-Receiver Encryption for many to one communication, and Multiple-Receiver Encryption for many to many communication within the group of conspiring users. We then show that Anamorphic Encryption can be embedded in the randomness used in the encryption, and we give families of constructions that can be applied to numerous ciphers. In total the families cover classical encryption schemes, some of which in actual use. Among our examples is an anamorphic channel with much higher capacity than the regular channel. In sum, the work shows the very large extent of the potential futility of control and censorship over the use of strong encryption by the dictator (typical for and even stronger than governments engaging in the ongoing crypto-wars): While such limitations obviously hurt utility which encryption typically brings to safety in computing systems, they essentially, are not helping the dictator. While the actual implications of what we show here and what it means in practice require further policy and legal analyses and perspectives, the technical aspects regarding the issues are clearly showing the futility of the war against Cryptography. (10.56553/popets-2023-0104)
  DOI : 10.56553/popets-2023-0104
• Model-based inexact graph matching on top of DNNs for semantic scene understanding
  
  • Chopin Jérémy
  • Fasquel Jean-Baptiste
  • Mouchère Harold
  • Dahyot Rozenn
  • Bloch Isabelle
  Computer Vision and Image Understanding, Elsevier, 2023, 235, pp.103744. Deep learning based pipelines for semantic segmentation often ignore structural information available on annotated images used for training. We propose a novel post-processing module enforcing structural knowledge about the objects of interest to improve segmentation results provided by deep learning. This module corresponds to a "many-to-one-or-none" inexact graph matching approach, and is formulated as a quadratic assignment problem. Our approach is compared to a CNN-based segmentation (for various CNN backbones) on two public datasets, one for face segmentation from 2D RGB images (FASSEG), and the other for brain segmentation from 3D MRIs (IBSR). Evaluations are performed using two types of structural information (distances and directional relations, , this choice being a hyper-parameter of our generic framework). On FASSEG data, results show that our module improves accuracy of the CNN by about 6.3% (the Hausdorff distance decreases from 22.11 to 20.71). On IBSR data, the improvement is of 51% (the Hausdorff distance decreases from 11.01 to 5.4). In addition, our approach is shown to be resilient to small training datasets that often limit the performance of deep learning methods: the improvement increases as the size of the training dataset decreases. (10.1016/j.cviu.2023.103744)
  DOI : 10.1016/j.cviu.2023.103744
• Biohybrid Devices: Prototyping Interactive Devices with Growable Materials
  
  • Nicolae Madalina
  • Roussel Vivien
  • Koelle Marion
  • Huron Samuel
  • Steimle Jürgen
  • Teyssier Marc
  , 2023, pp.1-15. Living bio-materials are increasingly used in HCI for fabricating objects by growing. However, how to integrate electronics to make these objects interactive still needs to be clarified. This paper presents an exploration of the fabrication design space of Biohybrid Interactive Devices, a class of interactive devices fabricated by merging electronic components and living organisms. From the exploration of this space using bacterial cellulose, we outline a fabrication framework centered on the biomaterials‘ life cycle phases. We introduce a set of novel fabrication techniques for embedding conductive elements, sensors, and output components through biological (e.g. bio-fabrication and bio-assembling) and digital processes. We demonstrate the combinatory aspect of the framework by realizing three tangible, wearable, and shape-changing interfaces. Finally, we discuss the sustainability of our approach, its limitations, and the implications for bio-hybrid systems in HCI. (10.1145/3586183.3606774)
  DOI : 10.1145/3586183.3606774
• Fighting selection bias in statistical learning: application to visual recognition from biased image databases
  
  • Clémençon Stéphan
  • Laforgue Pierre
  • Vogel Robin
  Journal of Nonparametric Statistics, American Statistical Association, 2023, 36 (3), pp.780-803. In practice, and especially when training deep neural networks, visual recognition rules are often learned based on various sources of information. On the other hand, the recent deployment of facial recognition systems with uneven performances on different population segments has highlighted the representativeness issues induced by a naive aggregation of the datasets. In this paper, we show how biasing models can remedy these problems. Based on the (approximate) knowledge of the biasing mechanisms at work, our approach consists in reweighting the observations, so as to form a nearly debiased estimator of the target distribution. One key condition is that the supports of the biased distributions must partly overlap, and cover the support of the target distribution. In order to meet this requirement in practice, we propose to use a low dimensional image representation, shared across the image databases. Finally, we provide numerical experiments highlighting the relevance of our approach (10.1080/10485252.2023.2259011)
  DOI : 10.1080/10485252.2023.2259011
• Watt-level Holmium-doped Fibre Amplifiers Pumped by Broadband Thulium-doped ASE Sources at 1860 nm
  
  • Delavaux Jean-Marc
  • Tench Robert E
  • Singh Save
  • Amavigan Alexandre
  • Walasik Wiktor
  • Jaouën Yves
  , 2023. We report the design and demonstration of novel 2050 nm Watt-level Ho-doped fibre amplifiers that are pumped with broad spectrum Watt-level Tm-doped ASE sources centred at 1860 nm instead of conventional narrow linewidth semiconductor or fibre laser sources. Our approach is simpler and more cost effective than the standard laser based pumping means, and leads to similar amplifier performance.
• Reasoning about Intuitionistic Computation Tree Logic
  
  • Catta Davide
  • Murano Aniello
  • Malvone Vadim
  , 2023, 391, pp.42-48. In this paper, we define an intuitionistic version of Computation Tree Logic. After explaining the semantic features of intuitionistic logic, we examine how these characteristics can be interesting for formal verification purposes. Subsequently, we define the syntax and semantics of our intuitionistic version of CTL and study some simple properties of the so obtained logic. We conclude by demonstrating that some fixed-point axioms of CTL are not valid in the intuitionistic version of CTL we have defined. (10.4204/EPTCS.391.6)
  DOI : 10.4204/EPTCS.391.6
• Intrinsic weaknesses of IDSs to malicious adversarial attacks and their mitigation
  
  • Chaitou Hassan
  • Robert Thomas
  • Leneutre Jean
  • Pautet Laurent
  Communications in Computer and Information Science, Springer Verlag, 2023, 1849, pp.122-155. Intrusion Detection Systems (IDS) are essential tools to protect network security from malicious traffic. IDS have recently made significant advancements in their detection capabilities through deep learning algorithms compared to conventional approaches. However, these algorithms are vulnerable to meta-attacks, also known as adversarial evasion attacks, which are attacks that improve already existing attacks, specifically their ability to evade detection. Deep learning-based IDS, in particular, are particularly susceptible to adversarial evasion attacks that use Generative Adversarial Networks (GAN). Nonetheless, well-known strategies have been proposed to cope with this threat. However, these countermeasures lack robustness and predictability, and their performance can be either remarkable or poor. Such robustness issues have been identified even without adversarial evasion attacks, and mitigation strategies have been provided. This paper identifies and formalizes threats to the robustness of IDSs against adversarial evasion attacks. These threats are enabled by flaws in the dataset's structure and content rather than its representativeness. In addition, we propose a method for enhancing the performance of adversarial training by directing it to focus on the best evasion candidates samples within a dataset. We find that GAN adversarial attack evasion capabilities are significantly reduced when our method is used to strengthen the IDS. (10.1007/978-3-031-45137-9_6)
  DOI : 10.1007/978-3-031-45137-9_6
• From Probabilistic Programming to Complexity-based Programming
  
  • Sileno Giovanni
  • Dessalles Jean-Louis
  , 2024, 1948, pp.304-317. The paper presents the main characteristics and a preliminary implementation of a novel computational framework named Com-pLog. Inspired by probabilistic programming systems like ProbLog, Com-pLog builds upon the inferential mechanisms proposed by Simplicity Theory, relying on the computation of two Kolmogorov complexities (here implemented as min-path searches via ASP programs) rather than probabilistic inference. The proposed system enables users to compute ex-post and ex-ante measures of unexpectedness of a certain situation, mapping respectively to posterior and prior subjective probabilities. The computation is based on the specification of world and mental models by means of causal and descriptive relations between predicates weighted by complexity. The paper illustrates a few examples of application: generating relevant descriptions, and providing alternative approaches to disjunction and to negation. (10.1007/978-3-031-50485-3_32)
  DOI : 10.1007/978-3-031-50485-3_32
• Obstruction Logic: A Strategic Temporal Logic to Reason About Dynamic Game Models
  
  • Catta Davide
  • Leneutre Jean
  • Malvone Vadim
  , 2023, 372, pp.365 - 372. Games that are played in a dynamic model have been studied in several contexts, such as cybersecurity and planning. In this paper, we introduce a logic for reasoning about a particular class of games with temporal goals played in a dynamic model. In such games, the actions of a player can modify the game model itself. We show that the model-checking problem for our logic is decidable in polynomial-time. Then, using this logic, we show how to express interesting properties of cybersecurity games defined on attack graphs. (10.3233/FAIA230292)
  DOI : 10.3233/FAIA230292
• The Impact of Strategies and Information in Model Checking for Multi-Agent Systems
  
  • Malvone Vadim
  , 2023, 391, pp.63-70. System correctness is one of the most crucial and challenging objectives in software and hardware systems. With the increasing evolution of connected and distributed systems, ensuring their correctness requires the use of formal verification for multi-agent systems. In this paper, we present a summary of certain results on model checking for multi-agent systems that derive from the selection of strategies and information for agents. Additionally, we discuss some open directions for future research. (10.4204/EPTCS.391.8)
  DOI : 10.4204/EPTCS.391.8
• Post-hoc Explainable AI for Black Box Models on Tabular Data
  
  • Radulovic Nedeljko
  , 2023. Current state-of-the-art Artificial Intelligence (AI) models have been proven to be verysuccessful in solving various tasks, such as classification, regression, Natural Language Processing(NLP), and image processing. The resources that we have at our hands today allow us to trainvery complex AI models to solve different problems in almost any field: medicine, finance, justice,transportation, forecast, etc. With the popularity and widespread use of the AI models, the need toensure the trust in them also grew. Complex as they come today, these AI models are impossible to be interpreted and understood by humans. In this thesis, we focus on the specific area of research, namely Explainable Artificial Intelligence (xAI), that aims to provide the approaches to interpret the complex AI models and explain their decisions. We present two approaches STACI and BELLA which focus on classification and regression tasks, respectively, for tabular data. Both methods are deterministic model-agnostic post-hoc approaches, which means that they can be applied to any black-box model after its creation. In this way, interpretability presents an added value without the need to compromise on black-box model's performance. Our methods provide accurate, simple and general interpretations of both the whole black-box model and its individual predictions. We confirmed their high performance through extensive experiments and a user study.
• Reflection sensitivity of dual-state quantum dot lasers
  
  • Jin Zhiyong
  • Huang Heming
  • Zhou Yueguang
  • Zhao Shiyuan
  • Ding Shihao
  • Yao Yong
  • Wang Cheng
  • Xu Xiaochuan
  • Grillot Frédéric
  • Duan Jianan
  Photonics research, Optical Society of America, 2023, 11 (10), pp.1713-1722. This work experimentally and theoretically demonstrates the effect of excited state lasing on the reflection sensitivity of dual-state quantum dot lasers, showing that the laser exhibits higher sensitivity to external optical feedback when reaching the excited state lasing threshold. This sensitivity can be degraded by increasing the excited-to-ground-state energy separation, which results in a high excited-to-ground-state threshold ratio. In addition, the occurrence of excited state lasing decreases the damping factor and increases the linewidth enhancement factor, which leads to a low critical feedback level. These findings illuminate a path to fabricate reflection-insensitive quantum dot lasers for isolator-free photonic integrated circuits. (10.1364/prj.494393)
  DOI : 10.1364/prj.494393
• Exploring Physical Latent Spaces for High-Resolution Flow Restoration
  
  • Paliard Chloé
  • Thuerey Nils
  • Um Kiwon
  , 2023. We explore training deep neural network models in conjunction with physics simulations via partial differential equations (PDEs), using the simulated degrees of freedom as latent space for a neural network. In contrast to previous work, this paper treats the degrees of freedom of the simulated space purely as tools to be used by the neural network. We demonstrate this concept for learning reduced representations, as it is extremely challenging to faithfully preserve correct solutions over long time-spans with traditional reduced representations, particularly for solutions with large amounts of small scale features. This work focuses on the use of such physical, reduced latent space for the restoration of fine simulations, by training models that can modify the content of the reduced physical states as much as needed to best satisfy the learning objective. This autonomy allows the neural networks to discover alternate dynamics that significantly improve the performance in the given tasks. We demonstrate this concept for various fluid flows ranging from different turbulence scenarios to rising smoke plumes. (10.2312/vmv.20231243)
  DOI : 10.2312/vmv.20231243