Francis Rousseaux - Academia.edu (original) (raw)

Papers by Francis Rousseaux

Lecture Notes in Computer Science, 2023

HAL (Le Centre pour la Communication Scientifique Directe), 2017

http://www.univ-reims.fr Résumé. A partir des premières hypothèses de travail construites pour la... more http://www.univ-reims.fr Résumé. A partir des premières hypothèses de travail construites pour la projet Exploration chronotopique d'un territoire parisien, l'équipe de recherche s'est élargie à de nouvelles disciplines apparues comme nécessaires pour explorer les matériaux identifiées et les interpréter avec des outils adaptés. Deux terrains sont menés conjointement : une ethnographie visant à comprendre comment se constitue l'expérience du temps dans la ville contemporaine, et une approche « objectivante » cherchant à identifier les données et les indicateurs pertinents pour révéler les assises chronotopiques sous-jacentes à certaines expériences du temps vécu dans les lieux habités. Les données horaires et calendaires recueillies au cours d'une première phase de recherche ont pu être structurées en un modèle de base de données permettant d'une part un traitement informatique visant à repérer des clusters horaires et calendaires et, dans un deuxième temps, à interpréter, à travers une approche géomatique, ces clusters dans l'espace de la ville pour en affiner la lecture et construire une analyse croisée avec les résultats de l'ethnographie.

Frontiers in Artificial Intelligence

The Bidirectional Encoder Representations from Transformers (BERT) architecture offers a cutting-... more The Bidirectional Encoder Representations from Transformers (BERT) architecture offers a cutting-edge approach to Natural Language Processing. It involves two steps: 1) pre-training a language model to extract contextualized features and 2) fine-tuning for specific downstream tasks. Although pre-trained language models (PLMs) have been successful in various text-mining applications, challenges remain, particularly in areas with limited labeled data such as plant health hazard detection from individuals' observations. To address this challenge, we propose to combine GAN-BERT, a model that extends the fine-tuning process with unlabeled data through a Generative Adversarial Network (GAN), with ChouBERT, a domain-specific PLM. Our results show that GAN-BERT outperforms traditional fine-tuning in multiple text classification tasks. In this paper, we examine the impact of further pre-training on the GAN-BERT model. We experiment with different hyper parameters to determine the best co...

Lecture notes in business information processing, 2022

To fulfil the increasing need for food of the growing population and face climate change, modern ... more To fulfil the increasing need for food of the growing population and face climate change, modern technologies have been applied to improve different farming processes. One important application scenario is to detect and measure natural hazards using sensors and data analysis techniques. Crowdsensing is a sensing paradigm that empowers ordinary people to contribute with data their sensor-enhanced mobile devices gather or generate. In this paper, we propose to use Twitter as an open crowdsensing platform for acquiring farmers knowledge. We proved this concept by applying pre-trained language models to detect individual's observation from tweets for pest monitoring.

2016 IEEE International Conference on Communications (ICC), 2016

Nowadays, wireless communication technologies have high influence on our daily lives. The vehicul... more Nowadays, wireless communication technologies have high influence on our daily lives. The vehicular ad hoc networks (VANETs) is now evolving quickly and attract attention from road operators, car manufacturers and governments. We unveil a fine traffic flow optimization method with VANETs. On the macroscopic level, our model identifies road segment profiles. On the microscopic level, vehicles send and receive alerts using low-level-of-details data. This paper gives workflows for both vehicle and infrastructure sides.

Proceedings of the 9th International Conference on Management of Digital EcoSystems, 2017

Music Information Retrieval (MIR) is a fifteen years old Web Intelligence community, gathering re... more Music Information Retrieval (MIR) is a fifteen years old Web Intelligence community, gathering researchers, practitioners, artists and music consumers around music, digital technologies, computer science, signal processing and new services. Faced with worldwide tensions over personal data legislation (after sensational legal proceedings and difficult transatlantic negotiations, the Safe Harbor system linking the EU and the USA was replaced by Privacy Shield on February 2, 2016). At the same time the European General Data Protection Regulation (GDPR) was adopted on April 14 2016 and will apply on May 25, 2018. The MIR community must consider one aspect of its international vocation to contribute to the development of scientific research in innovative uses of sound and digital music, by helping its members familiarize themselves with the right to privacy and making Music Information Retrieval researchers aware of the methods they can use to control risks linked to development and prototype testing, but also by helping to make the expert debate clearer for a better balance at the international level between scientific research, current practices and the law.

— Object-oriented computer science was created to simulate our activities of placing objects in i... more — Object-oriented computer science was created to simulate our activities of placing objects in identified and labeled class structures. As we know, its success was immediate. Recently, an innovative trend appeared. It is characterized by the mobilization of object-oriented computer science for the organization of our collections which are considered like heaps of objects waiting to be classified in ad-hoc classes that could be created at the same time. Undeniably, collecting is an older activity than classifying, in so far as it allows the useful experimentation of the concepts of extension (in the case of a spatiotemporal, even temporary and ephemeral arrangement) and of intension in the idea of an abstract order of similarities. We always put together a collection of something, which makes it impossible to typify the activity regardless of the objects, and which therefore disturbs the modeler's customary practices.

Les plateformes de travail e-collaboratif permettent aux personnes travaillant sur des sujets sim... more Les plateformes de travail e-collaboratif permettent aux personnes travaillant sur des sujets similaires de partager un espace de travail, d’echanger du contenu et de faire des recherches « intelligentes ». Cela necessite un investissement personnel important de la part des utilisateurs qui seuls pourraient frei- ner les usages individualistes. Cet article propose une premiere reflexion pour solutionner le probleme de manque de la motivation des utilisateurs dans le travail collaboratif numerique et sur ce que l’intelligence artificielle peut apporter au travail du futur soutenu par des portails e-collaboratifs. Apres un etat de l’art des theories de la psychologie positive, nous mettons l’accent sur les facteurs motivationnels a prendre en compte dans la personnalisation d’un ecosysteme digital intrinsequement motivants pour un collectif d’utilisateur afin de promouvoir la creation participative de connaissances et l’innovation collaborative.

Intelligent Systems Reference Library, 2017

Recently, studies about culturally-intelligent systems have arisen to manage digitized cultural d... more Recently, studies about culturally-intelligent systems have arisen to manage digitized cultural diversity. The current systems possess an artificial awareness of cultures by mediating them through representations. Coming from an etic approach, these universal representations facilitate the mediation of different cultures but limit their understanding and thus, prevent the development of an higher degree of awareness. In this research, we propose a methodology to construct artificial cultural awareness from emic-based representations. We tested the latter through an experiment on the domain of 'abortion' with the Pro-Choice and Pro-Life communities.

2017 4th International Conference on Control, Decision and Information Technologies (CoDIT), 2017

Today, the development of the semantic web is heavily impacted by the knowledge acquisition bottl... more Today, the development of the semantic web is heavily impacted by the knowledge acquisition bottleneck. To address this problem, the discovery of conceptual structures have to become automatic. In addition, it has to guarantee that these structures are socially-constructed and consensually-shared in accordance with the requirements for designing ontologies. In this research, we present a process to automatically discover ready-made cultural knowledge structures from texts fitting the needs of ontology engineering. Relying on a framework coming from cognitive anthropology, we designed the latter to identify hypernym/hyponym relations. During our experiment we obtained a promising 92.46% precision.

HAL (Le Centre pour la Communication Scientifique Directe), 2012

La préservation des oeuvres musicales était ces derniers siècles conditionnée par le recours à un... more La préservation des oeuvres musicales était ces derniers siècles conditionnée par le recours à un substrat culturel stable, c'est-à-dire la partition, une organologie bien identifiée avec la classification des instruments et la lutherie qui permettait de fabriquer ces instruments. L'apparition des systèmes électroniques et numériques a chamboulé le monde de la composition, lui faisant perdre ses traditions pluri-séculaires et mettant en péril la préservation de la musique contemporaine. Devant la nécessité de mettre à jour les oeuvres pour lutter contre l'obsolescence technologique, il nous paraît très intéressant d'étudier les processus de composition desquels nous pouvons extraire des informations pertinentes, nécessaires pour ressaisir certaines intentions originales du compositeur. Notre approche se fonde sur la création d'un modèle capable de représenter ce processus de production qui sera utilisé dans le projet ANR Gamelan au sein d'un ensemble d'outils permettant de le visualiser et de l'étudier. Nos premiers résultats suite à une phase d'immersion dans les pratiques actuelles de composition concernent la représentation de certaines informations extraites d'une production et qu'il nous semble déjà vital de posséder lors d'une migration d'une oeuvre afin d'en préserver l'authenticité.

Apart from the traditional uses of computing in the fields of music and musicology, human-compute... more Apart from the traditional uses of computing in the fields of music and musicology, human-computer systems are emerging in reaction to this tradition, which is already effective, though still young. Whereas these unique systems operate in the usual fields of music's expansion, they cannot be systematically referred to known musicological categories. On the contrary, the experiments made possible by these systems inaugurate new uses where listening, composing and music transmission are merged in a gesture sometimes described as "music-ripping". We will show how music-ripping practices provoke traditional musicology, whose canonical categories are powerless for our purpose. To achieve this purpose, we will:- give an explicit minimal set of categories sufficient to underlie the usual models of computer assisted music;- do the same thing for human-computer (anti-musicological?) systems which existence is disturbing to us;- examine the conditions necessary to reduce the se...

Nous presentons ici les premiers resultats obtenus dans notre recherche de nouvelles formes d&#39... more Nous presentons ici les premiers resultats obtenus dans notre recherche de nouvelles formes d'operas sur supports informatiques, recherche conjuguee avec la creation d'une oeuvre de ce genre. La cinematique est articulee selon un modele informatique de forme ouverte issu des conceptions mises en oeuvre par Boucourechliev dans « Archipel IV », ainsi que des menus integres au flux dramatique. Plus que des contenus narratifs ou textuels, il s'agit dans notre oeuvre de mettre en scene des fragments musicaux, en representant leur deploiement par des couloirs en trois dimensions parcourus au tempo correspondant. Les caracteristiques de ces couloirs sont fondees sur le suivi des lignes de basse et de dessus, et les variations de densite des agregats verticaux.

HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific re... more HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

In recent years, MIR research has continued to focus more and more on user feedback, human subjec... more In recent years, MIR research has continued to focus more and more on user feedback, human subjects data, and other forms of personal information. Concurrently, the European Union has adopted new, stringent regulations to take effect in the coming years regarding how such information can be collected, stored and manipulated, with equally strict penalties for being found in violation of the law. Here, we provide a summary of these changes, consider how they relate to our data sources and research practices, and identify promising methodologies that may serve researchers well, both in order to be in compliance with the law and conduct more subject-friendly research. We additionally provide a case study of how such changes might affect a recent human subjects project on the topic of style, and conclude with a few recommendations for the near future. This paper is not intended to be legal advice: our personal legal interpretations are strictly mentioned for illustration purpose, and reader should seek proper legal counsel.

Bridging People and Sound, 2017

Halfway between music analysis and graph visualization, we propose tonal pitch representations fr... more Halfway between music analysis and graph visualization, we propose tonal pitch representations from the chromatic scale. On the first part of the experiment, a 12-node graph is connected as a Rhythm Network and visualized with a Circular Layout, commonly known as Pitch constellation. This particular graph topology focuses on node structure and gives strength to weak edges. At this occasion, we unveil the Singularity Threshold, giving an opportunity to isolate structure from singular parts of melodies. Where usual Pitch constellations focus on chords, we focus on successive pitch intervals. On the second part, we propose a rhythm representation using a Force-Directed layout. In addition to structural information, our second technique shows proximal and peripheral elements. This experiment features 6 melodies that we propose to visualize using Gephi.

Proceedings of the 9th International Joint Conference on Computational Intelligence, 2017

Digital knowledge gave birth to massive communication spaces, new access paths and new cleavages.... more Digital knowledge gave birth to massive communication spaces, new access paths and new cleavages. Our experiment deals with the challenging issue of accessing this knowledge on the Internet. Computer scientists set up prediction algorithms and recommender engines. This way, knowledge access is partly automatized. Using a real-life dataset, our goal is to simulate the iterative behavior shift produced by most used recommender engines. On this basis, we show that in the context of recommendation, existing evaluation metrics are driven by prediction testing methods and we argue that ambiguity has to be raised between prediction and recommendation. Secondly, we propose alternative evaluation metrics for recommendation systems, targeting mitigating the bias problem of information cascades and confirmation biases.

2016 Intl IEEE Conferences on Ubiquitous Intelligence & Computing, Advanced and Trusted Computing, Scalable Computing and Communications, Cloud and Big Data Computing, Internet of People, and Smart World Congress (UIC/ATC/ScalCom/CBDCom/IoP/SmartWorld), 2016

At the same time as Big Data technologies are being constantly refined, the legislation relating ... more At the same time as Big Data technologies are being constantly refined, the legislation relating to data privacy is changing. The invalidation by the Court of Justice of the European Union on October 6, 2015, of the agreement known as "Safe Harbor", negotiated by the European Commission on behalf of the European Union with the United States has two consequences. The first is to announce its replacement by a new, still fragile, program, the "Privacy Shield", which isn't yet definitive, which could also later be repealed by the Court of Justice of the European Union. For example, we are expecting to hear the opinion in mid-April 2016 of the group of data protection authorities for the various states of the European Union, known as G29. The second is to mobilize the Big Data community to take control of the question of data privacy management, to put in place an adequate internal program.

2012 Third International Conference on Emerging Intelligent Data and Web Technologies, 2012