Dissertations / Theses on the topic 'Explicabilité de l'intelligence artificielle'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic 'Explicabilité de l'intelligence artificielle.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
Ben, Gamra Siwar. "Contribution à la mise en place de réseaux profonds pour l'étude de tableaux par le biais de l'explicabilité : Application au style Tenebrisme ("clair-obscur")." Electronic Thesis or Diss., Littoral, 2023. http://www.theses.fr/2023DUNK0695.
Face detection from Tenebrism paintings is of growing interest to art historians and researchers in order to estimate the illuminant location, and thereby answer several technical questions. Deep learning is gaining increasing interest due to is high performance capabilities. An optimization of Faster Region based Convolutional Neural Network has demonstrated its ability to effectively address challenges and deliver promising face detection results from Tenebrism paintings. However, deep neural networks are often characterized as "black box" because of the inherent complexity and non-linearity of their architectures. To tackle these issues, eXplainable Artificial Intelligence (XAI) is becoming an active researcj area to understand deep models. So, we propose a novel iterative XAI method based on guided perturbations to explain model's application
Schmitt, Michel. "Des Algorithmes morphologiques à l'intelligence artificielle." Phd thesis, École Nationale Supérieure des Mines de Paris, 1989. http://pastel.archives-ouvertes.fr/pastel-00001572.
Schmitt, Michel. "Des Algorithmes morphologiques à l'intelligence artificielle." Grenoble 2 : ANRT, 1988. http://catalogue.bnf.fr/ark:/12148/cb37618487h.
Merabet, Samir. "Vers un droit de l'intelligence artificielle." Thesis, Aix-Marseille, 2018. http://www.theses.fr/2018AIXM0528.
Even if its appearance is recent among technological inventions history, artificial intelligence has nevertheless quickly established itself, disrupting economy and the job market. Yet, upon assessment, it seems that these two forms of intelligence cannot be regarded as equivalent. Even if artificial intelligence borrows some aspects of human intelligence, many others are missing. Conscience, reason and emotions are unknown to machines, even intelligent ones. Yet, law rests upon such qualities. Hence, applying rules created for human to intelligent computer systems may be inappropriate. Indeed, the confrontation between law and artificial intelligence reveals the existence of a paradigm on which positive law is based. To a large extent, French law relies on the subjectivity proper to humans. All branches of law appear to be concerned, civil law as well as criminal law or intellectual property law. Therefore, the legal regime of artificial intelligence seems very uncertain. Consequently, the purpose of this study is to clear up the doubts surrounding the nature of artificial intelligence in order to neatly distinguish it from human intelligence. Eventually, the acknowledgment of the fundamental difference opposing these two forms of intelligence should lead to the recognition of a new public order of humanity and the preservation of an exclusive field for human intelligence
Merabet, Samir. "Vers un droit de l'intelligence artificielle." Electronic Thesis or Diss., Aix-Marseille, 2018. https://buadistant.univ-angers.fr/login?url=https://bibliotheque.lefebvre-dalloz.fr/secure/isbn/9782247201235.
Even if its appearance is recent among technological inventions history, artificial intelligence has nevertheless quickly established itself, disrupting economy and the job market. Yet, upon assessment, it seems that these two forms of intelligence cannot be regarded as equivalent. Even if artificial intelligence borrows some aspects of human intelligence, many others are missing. Conscience, reason and emotions are unknown to machines, even intelligent ones. Yet, law rests upon such qualities. Hence, applying rules created for human to intelligent computer systems may be inappropriate. Indeed, the confrontation between law and artificial intelligence reveals the existence of a paradigm on which positive law is based. To a large extent, French law relies on the subjectivity proper to humans. All branches of law appear to be concerned, civil law as well as criminal law or intellectual property law. Therefore, the legal regime of artificial intelligence seems very uncertain. Consequently, the purpose of this study is to clear up the doubts surrounding the nature of artificial intelligence in order to neatly distinguish it from human intelligence. Eventually, the acknowledgment of the fundamental difference opposing these two forms of intelligence should lead to the recognition of a new public order of humanity and the preservation of an exclusive field for human intelligence
Farcis, Louis. "Cellule MRAM neuromorphique pour l'Intelligence Artificielle." Electronic Thesis or Diss., Université Grenoble Alpes, 2024. http://www.theses.fr/2024GRALT015.
The rapid evolution of the society in term computational needs tends to target application that were once human exclusive. Development of deep learning algorithm made possible overpassing some capabilities in the task the human brain were performing the best as for instance image and speech recognition, decision making and optimization problems. However, The Artificial Neural Networks (ANNs) capable of solving these tasks use intensively massive amount of data and requires multiple operations at the same time. When running ANN algorithms in a classical Von Neumann architecture where the computing unit is separated by the memory, latency and high energy consumption starts to rise exponentially with the size of the emulated neural network. From this observation, the scientific community starts looking at brain inspired computing sheme to overcome the current limitation. In particular, spiking neural networks (SNNs) were early predicted by W. Maass in 1997 to be a suitable candidate to leverage the sparsity of the network while showing egal if not better results than the first generations of ANN. Up until now, some Application Specific Integrated Circuits (ASICs) were proposed by company such as Intel and IBM to emulate SNN with CMOS-based based technology. The number of transistors needed to accomplish some critical functionality like spiking neurons in these solution is still very large and not suitable for downscaling strategies. In this context, new hardware solutions were proposed to emulate the synaptic and neuronal features while reducing the footprint and the energy consumption. In particular, various types of nano-synapses based on emerging non-volatile memory (NVM) explored multi-level synaptic weights and short-term/long-term memory. Among them, the spintronic solutions are the most advanced in maturity compared to the other technology because magnetic-random-access-memory technology, which represents a binary synapse, has already reached the market for ten years. However, a spiking neuron compatible with spintronic-based synapse is still missing in the literature.The thesis takes place in this context of developing new solutions with spintronics in order to emulate spiking neurons. Magnetic tunnel Junctions (MTJs) have been widely used in spintronics as memory unit because of their high endurance while demonstrating relatively small energy for writing and reading operation and are BEOL CMOS compatible. The solution elaborate along the manuscript takes all the benefits of the MTJs to design a spiking neurons based on the windmill dynamics.The dual-free layer MTJ concept is modelled, designed, nano-patterned and electrically characterized to give a constructive outlook on how viable is this structure for emulation spiking neuron
Ringuet, Jean-Noël. "L'intelligence artificielle : reproduction ou simulation de l'esprit." Thesis, University of Ottawa (Canada), 1988. http://hdl.handle.net/10393/5502.
Pistilli, Giada. "Pour une éthique de l'intelligence artificielle conversationnelle." Electronic Thesis or Diss., Sorbonne université, 2024. http://www.theses.fr/2024SORUL038.
This research aims to probe the ethical intricacies of conversational Artificial Intelligence (AI), specifically focusing on Large Language Models and conversational agents. This manuscript constructs a framework that melds empirical analysis with philosophical discourse. We aim to urgently advocate for a well-founded ethical structure for conversational AI, highlighting the necessity to involve all stakeholders, from developers to end-users. Firstly, we champion the integration of engineering and other scientific disciplines with philosophy, facilitating a more nuanced understanding of the ethical dimensions underpinning AI. This collaborative approach allows for a richer, more informed ethical discourse. Secondly, we advocate for the dynamic use of applied ethical frameworks as foundational guides for setting the initial objectives of an AI system. These frameworks serve as evolving tools that adapt to the ethical complexities encountered during development and deployment. Lastly, grounded in hands-on, interdisciplinary research, we make an argument for the prioritization of narrow, task-specific AI over Artificial General Intelligence, a stance that is based on the enhanced feasibility of ethical oversight and technical controllability.With this research, we aim to contribute to the literature on AI ethics, enriching the academic discourse in both philosophy and computer science
Voyiatzis, Konstantinos. "Utilisation de l'intelligence artificielle pour les problèmes d'ordonnancement." Paris 9, 1987. https://portail.bu.dauphine.fr/fileviewer/index.php?doc=1987PA090102.
The present work concerns the scheduling of the products in a production system from "jobshop" type. This problem is a combinatorial problem and has an incompatible solution with the "real time" contraint. We demonstrate in which way this approach "artificial intelligence" can conduct to an acceptable scheduling in real time. The first chapter presents the structure of the artificial memory we use. The second one tells us an example of utilization on this approach. The appendix i remains the basis of production managment. The appendix ii gives us the basis of the automatic classification used in severals parts of this work. And the appendix iii, contains the software of simulation written in slam
Laï, Claude. "Applications de l'intelligence artificielle à l'instrumentation en ligne." Aix-Marseille 2, 1987. http://www.theses.fr/1987AIX22024.
Laï, Claude. "Application de l'intelligence artificielle à l'instrumentation en ligne." Grenoble 2 : ANRT, 1987. http://catalogue.bnf.fr/ark:/12148/cb376068857.
Voyiatzis, Konstantinos. "Utilisation de l'intelligence artificielle pour les problèmes d'ordonnancement." Grenoble 2 : ANRT, 1987. http://catalogue.bnf.fr/ark:/12148/cb376106812.
TOUCHAIS, ROLAND. "Enseignement assiste par ordinateur des algorithmes de l'intelligence artificielle." Paris 11, 1992. http://www.theses.fr/1992PA112024.
PERSON, PATRICK. "Les reseaux bayesiens : un nouvel outil de l'intelligence artificielle." Paris 6, 1991. http://www.theses.fr/1991PA066607.
Rommel, Yves. "Apport de l'intelligence artificielle à l'aide à la décision multicritère." Paris 9, 1989. https://portail.bu.dauphine.fr/fileviewer/index.php?doc=1989PA090004.
The aim of this thesis is to show the contribution of artificial intelligence technics (AI) to multicriteria decision making. The AI technics which have been used concern knowledge representation technics in a declarative manner, deductive technics of problem resolution and also "inductive" technics of machine learning. Research has been developped into two opposite directions relatively to the modelisation approach as follows: -the first direction consists of defining how much AI contributes to the traditional multicriteria analysis frame i. E. We consider here the multicriteria problems using a scores table and according to the three operational approaches to the aggregation of preferences (synthesis criterion, outranking and interactive procedure); - the second direction consists of considering multicriteria methodology within the expert-system architecture and modelisation framework. This rests on the fact that expert-systems naturally deal with multicriteria problems
Gaborieau, Patrice. "Le sens comme tâche ou Quand l'intelligence se fait artificielle." Rennes 2, 2001. http://www.theses.fr/2001REN20033.
Generative grammar causes problems as a theory of language. Working on the same lines as positivism, it does not consider the anthropological circularity of linguistic phenomenon : deconstructing its scientific object reveals that it comes under the heading of separate principles of human rationality. Galilean epistemology, to which it appeals, however, gives a way to show off its point of consistensy. As a matter of facts, Galilean nature sciences, as they appeared in the 17th century, are distinguished by their technical immanent structure, linked to measuring apparatus and algebra. It is therefore possible to formulate the assumption that generative grammar is a technically equipped 'production' of language, before being an explanatory theory
Neggaz, Mohamed Ayoub. "Accélérateurs Matériels pour l'Intelligence Artificielle. Etude de Cas : Voitures Autonomes." Thesis, Valenciennes, Université Polytechnique Hauts-de-France, 2020. http://www.theses.fr/2020UPHF0017.
Since the early days of the DARPA challenge, the design of self-driving cars is catching increasing interest. This interest is growing even more with the recent successes of Machine Learning algorithms in perception tasks. While the accuracy of thesealgorithms is irreplaceable, it is very challenging to harness their potential. Realtime constraints as well as reliability issues heighten the burden of designing efficient platforms.We discuss the different implementations and optimization techniques in this work. We tackle the problem of these accelerators from two perspectives: performance and reliability. We propose two acceleration techniques that optimize time and resource usage. On reliability, we study the resilience of Machine Learning algorithms. We propose a tool that gives insights whether these algorithms are reliable enough forsafety critical systems or not. The Resistive Associative Processor accelerator achieves high performance due to its in-memory design which remedies the memory bottleneck present in most Machine Learning algorithms. As for the constant multiplication approach, we opened the door for a new category of optimizations by designing instance specific accelerators. The obtained results outperforms the most recent techniques in terms of execution time and resource usage. Combined with the reliability study we conducted, safety-critical systems can profit from these accelerators without compromising its security
Vinurel, Jean-Jacques. "Une application de l'intelligence artificielle à l'enseignement assisté par ordinateur." Paris 6, 1986. http://www.theses.fr/1986PA066379.
Vinurel, Jean-Jacques. "Une Application de l'intelligence artificielle à l'enseignement assisté par ordinateur." Grenoble 2 : ANRT, 1986. http://catalogue.bnf.fr/ark:/12148/cb376017298.
Souza, Marco Antonio Lucas de. "Intelligence artificielle et philosophie : les critiques de H. L. Dreyfus et J. Searle à l'intelligence artificielle." Thèse, Université du Québec à Trois-Rivières, 1992. http://depot-e.uqtr.ca/5367/1/000597800.pdf.
Dubois, Gilles. "Apport de l'intelligence artificielle à la coopération de systèmes d'information automatisée." Lyon 3, 1997. http://www.theses.fr/1997LYO33004.
Recent advances in distributed systems, computer network and database technology have changed the information processing needs of organizations. Current information systems should integrate various heterogeneous sources of data and knowledge according to distributed logical and physical requirements. An automated information system is perceived as a set of autonomous components which work in a synergistic manner by exchanging information expertise and coordinating their activities. In order for this exchange to be judicious, the individual systems must agree on the meaning of their exchanged information to solve conflicts due to heterogeneity. We have chosen an object oriented model as canonical model. The object model overcomes component heterogeneity and respects the autonomy of local systems in a distributed context. The cooperation structure uses artificial intelligence techniques to solve both structural and semantic conflicts. A dynamic description of information sources deals with local evolution and is involved in global queries treatment. An extension of the proposal exploits agents interactions to bring cognitive capabilities to the cooperation structure. The contribution of multi-agent systems to information system cooperation is argued. Technical choices to implement a prototype in an object oriented environment are described
Wagner, Pierre. "Machine et pensée : l'importance philosophique de l'informatique et de l'intelligence artificielle." Paris 1, 1994. http://www.theses.fr/1994PA01A009.
The goal of this work is to value the philosophical importance of computer science and artificial intelligence, by studying the relations between machine and thought. The arguments raised for and against mechanism in the contemporary philosophy are examined successively. In this case, the theses of mechanical philosophy are based on computer science and on artificial intelligence. According to the strongest of these theses, a machine can think and a mechanical model of thought using ideas in computer science can be designed. With regard to artificial intelligence, the two main points examined are the following : fist, what can be expected from the progress made in this domain, by a philosophical investigation of the relations between machine and thought ? Second, is the parallel often drawn between philosophers as descartes, hobbes, leigniz or la mettrie on the one hand, and the representation of mind according to artificial intelligence on the other hand, a valid one ? On this last point, it is argued that the answer is no. As for computer science, it raises its own questions about machine and thought, questions that are often overlooked, and ignored by the debate on mechanism and artificial intelligence, while they no less deserve philosophical examination. These questions are raised by studying the relations between logic and computer science
Gaïti, Dominique. "L'utilisation des techniques de l'intelligence artificielle pour la gestion des reseaux." Paris 6, 1991. http://www.theses.fr/1991PA066493.
Claes, Gérard. "Contribution à l'application de l'intelligence artificielle pour l'enseignement assisté par ordinateur." Paris 11, 1988. http://www.theses.fr/1988PA112412.
Bourseau, Patrick. "Modelisation des connaissances de l'ingenieur de procedes : application de l'intelligence artificielle." Paris 6, 1993. http://www.theses.fr/1993PA066039.
Claes, Gérard. "Contribution à l'application de l'intelligence artificielle pour l'enseignement assisté par ordinateur." Grenoble 2 : ANRT, 1988. http://catalogue.bnf.fr/ark:/12148/cb376127000.
Neumann, Andreas. "Introduction d'outils de l'intelligence artificielle dans la prévision de pluie par radar." Phd thesis, Ecole Nationale des Ponts et Chaussées, 1991. http://tel.archives-ouvertes.fr/tel-00520834.
Racoceanu, Daniel. "Contribution à la Surveillance des Systèmes de Production en Utilisant l'Intelligence Artificielle." Habilitation à diriger des recherches, Université de Franche-Comté, 2006. http://tel.archives-ouvertes.fr/tel-00011708.
Nos contributions se sont orientées précisément vers la mise au point de systèmes de surveillance dynamiques intelligents, en abordant des problématiques liées à la détection dynamique et au diagnostic curatif et préventif. Les techniques de l'intelligence artificielle ont ainsi constituées la colonne vertébrale de nos études, avec des travaux allant de la conception théorique et le test de nouveaux outils de surveillance dynamique, jusqu'à la validation, le prototypage et l'exploitation industrielle des concepts développés.
Dans le domaine de la surveillance dynamique, une de nos contributions majeures correspond à l'introduction d'un réseau de neurones dynamique innovant, le Réseau Récurrent à base de Fonctions radiales (RRFR). En faisant appel aux propriétés dynamiques des architectures localement récurrentes, le RRFR se caractérise par une approche de reconnaissance locale – essentielle en surveillance industrielle –, tout en permettant – de part ses caractéristiques dynamiques intrinsèques – une détection précoce des paliers de dégradation et une robustesse vis-à-vis des fausses alarmes. L'existence d'une mémoire statique et d'une mémoire dynamique facilement paramétrables au sein de la même structure, confère au réseau une souplesse d'utilisation très intéressante, ainsi qu'un accès à des techniques d'apprentissage allégées. Par ailleurs, dans le but d'augmenter la rapidité et l'efficacité de l'apprentissage, une version améliorée de l'algorithme des k-moyennes, permet d'améliorer la robustesse des algorithmes et de se situer dans la zone optimale de « bonne généralisation ».
La prise en compte de l'incertain fait l'objet d'une deuxième partie de nos travaux, située dans la continuité de la première. Dans ce sens, l'utilisation de la logique floue en surveillance dynamique se montre très intéressante de part la proximité par rapport au raisonnement et à l'expérience des opérateurs et ingénieurs, ainsi que de part la possibilité d'assurer une traçabilité essentielle pour le retour d'expérience en maintenance. Un outil de type réseau de Petri flou dédié à la surveillance (RdPFS) des systèmes de production a ainsi constitué l'objet de cette étude. Cet outil, basé essentiellement sur la modélisation floue de la variable "instant d'apparition" de l'événement discret, est muni d'un mécanisme d'interfaçage intégré, inspiré des réseaux de Petri à synchronisations internes, permettant la communication entre les outils de modélisation, de surveillance et de reprise. Dans un tel système, l'utilisation de la logique floue offre une souplesse d'utilisation et une dimension prédictive très intéressante. Basé sur un modèle RdPFS construit à partir des règles logiques induites par l'arbre de défaillance du système surveillé, nous avons montré sa capacité d'analyse dynamique des défauts, en utilisant une approche temporelle floue.
Dans nos travaux les plus récents, les avantages des outils neuronaux dynamiques et de la logique floue ont été mis à l'œuvre conjointement dans le cadre de systèmes hybrides neuro-flous d'aide à la surveillance. L'outil hybride ainsi proposé comporte deux parties, une première pour la détection dynamique – utilisant le RRFR amélioré essentiellement au niveau des algorithmes d'apprentissage dynamiques – et une deuxième pour le diagnostic curatif et prédictif, utilisant un réseau neuro-flou construit à partir de l'arbre de défaillance et de l'AMDEC du système / sous-système surveillé. Il est à noter que ce type de système de surveillance dynamique utilise l'approche abductive de recherche de cause – élément indispensable pour un diagnostic efficace. Par ailleurs, l'apprentissage dynamique permet de commencer l'activité de surveillance dynamique même en l'absence de données très consistantes, en enrichissant et affinant les symptômes et les causes associées (respectivement les degrés de crédibilité de celles-ci) au fur et à mesure du fonctionnement du système, grâce aux algorithmes neuronaux incorporés.
Parmi les verrous technologiques abordés lors de nos contributions, une place importante est accordée à la surveillance dynamique par apprentissage en ligne, en traitant des problèmes liés à la rapidité et la flexibilité de l'apprentissage, au temps de réponse, au traitement de grands flux de données, ainsi qu'aux méthodes de validation utilisant le test sur des benchmarks et le prototypage industriel.
Dans ce sens, l'exploitation industrielle du réseau récurrent à fonctions de base radiales a représenté une étape importante de nos travaux. Elle nous a permis la mise en œuvre d'une application de surveillance dynamique en temps réel d'un système, en utilisant l'apprentissage dynamique distant via le web. Mettant en évidence l'intérêt des techniques développées dans la réorganisation de l'activité de surveillance grâce aux nouvelles technologies, cette exploitation industrielle a donné lieu à un brevet d'invention, déposé en collaboration avec une SSII bisontine.
La conception de l'outil de surveillance neuro-flou a été menée en utilisant des spécifications UML. Cette approche a été initiée (étude, normalisation, spécification) dans le cadre du projet Européen PROTEUS/ITEA et finalisée (prototypage) dans le cadre d'un projet financé par l'ANVAR. Essentiels pour l'industrialisation des systèmes proposés, l'information de départ est considérée sous un format industriel classique (ADD, AMDEC, ...) et la mise à jour est assurée par une réelle ouverture du système traduite par des liens permanents avec les systèmes d'acquisition (capteurs, SCADA, ...) et les systèmes de gestion (GMAO, ...).
Dans le domaine de la e-maintenance, un défi considérable est constitué par le besoin de normalisation des plates-formes de e-maintenance, dans l'objectif d'arriver à terme à une génération automatique de ces plateformes, et à un standard qui pourra constituer un guide pour tous les constructeurs d'équipement soucieux d'intégrer leur produit dans une plate-forme de ce type. Du point de vue scientifique, il s'agit d'arriver à une génération automatique ou interactive d'une telle plate-forme, en fonction des services participants et du contexte d'utilisation. L'objectif est très ambitieux, car générateur de nouveaux services et de nouvelles opportunités commerciales. Il constitue l'objectif principal du projet européen SHIVA, dans lequel se retrouvent les partenaires les plus actifs de PROTEUS, avec – mise a part des partenaires déjà existants comme Cegelec et Schneider – des partenaires industriels nouveaux, tout aussi prestigieux, comme la Division des Constructions Navales et Airbus.
Concernant les perspectives de nos travaux, s'inscrivant dans le cadre des mêmes tendances de décentralisation et de migration de l'intelligence vers les niveaux opérationnels, une direction d'étude intéressante est constituée par le domaine des réseaux de capteurs intelligents. Cette perspective pourra ainsi intégrer des tendances technologiques liées aux connecteurs intelligents (« smart connectors ») basés sur le multiplexage et sur les courants porteurs, afin de réduire le volume de câblage, des points de connexion et afin de mieux prendre en considération les contraintes de sûreté de fonctionnement de ce type de réseaux, contraintes qui – dans un avenir très proche – seront vraisemblablement déterminantes.
Par ailleurs, dans un contexte régional, lié à la création du pôle de compétitivité « microtechniques », la problématique de la fabrication des microsystèmes est une problématique présentant un intérêt croissant. En effet, de nombreux prototypes de microsystèmes sont proposés en phase de prototype, avec une création liée à une manière plutôt artisanale, tout à fait compréhensible – vu les technologies innovantes utilisées -, mais qui ne se prête pas à une commercialisation de ces produits. De ce fait, l'étude des microsystèmes de production semble être un domaine porteur et permettant de garder des emplois européens – souvent menacés par la délocalisation.
Enfin, le diagnostic à partir de l'imagerie médicale en utilisant les techniques de l'intelligence artificielle me semble un domaine où mes compétences en recherche (surveillance par IA) et en enseignement (traitement de l'image) peuvent se donner la main pour aborder une thématique nouvelle, présentant un grand intérêt dans le domaine des biotechnologies, dans lequel la France est bien engagée depuis un certain nombre d'années. Cette perspective correspond au projet de recherche propose dans le cadre de ma délégation CNRS au laboratoire IPAL – Image Perception, Access and Learning (actuellement FRE, UMI-CNRS a partir de 2006) de Singapour.
Lemaire, Alexis. "Application de l'hypercalculie et de l'informatique quantique gravifique à l'intelligence artificielle générale." Reims, 2010. http://www.theses.fr/2010REIMS020.
Lassègue, Jean. "L'intelligence artificielle et la question du continu : remarques sur le modèle de Turing." Phd thesis, Université de Nanterre - Paris X, 1994. http://tel.archives-ouvertes.fr/tel-00011541.
On décrit la notion de machine de Turing en tant que concept dans la théorie de la calculabilité. On étudie le contexte épistémologique dans lequel le concept est né dans les années trente : philosophiquement, la “querelle des fondements en mathématique”; mathématiquement, l'apparition des différents formalismes permettant de rendre compte de la calculabilité des fonctions, dont le formalisme de la machine de Turing.
On décrit dans la deuxième partie comment le concept de machine de Turing se transforme en modèle pour la théorie de la psychologie. La justification de cette transformation est étudiée à partir de l'expérience de pensée élaborée par Turing grâce au “jeu de l'imitation”. On interprète le sens de ce jeu d'un point de vue formaliste, probabiliste et psychologique. On finit par conclure à l'absence de “test de Turing” dans le jeu, contrairement à ce qui est cru généralement.
La troisième partie étudie la façon dont la notion de machine de Turing a servi de fondement à l'intelligence artificielle. Le modèle de Turing tel qu'il a été utilisé jusqu'à présent engendre deux types de théories dualistes de l'esprit : une théorie platonicienne et une théorie fonctionnaliste. On justifie une interprétation non-dualiste en mettant l'accent sur le rôle joué par le langage dans la constitution du modèle. On replace enfin le modèle dans une tradition historique plus large, qui va de C. Babbage à R. Thom.
Vidal, Jean-Philippe. "Assistance au calage de modèles numériques en hydraulique fluviale - Apports de l'intelligence artificielle." Phd thesis, Institut National Polytechnique de Toulouse - INPT, 2005. http://tel.archives-ouvertes.fr/tel-00010185.
Levasseur, Yan. "Techniques de l'intelligence artificielle pour la classification d'objets biologiques dans des images bidimensionnelles." Mémoire, École de technologie supérieure, 2008. http://espace.etsmtl.ca/115/1/LEVASSEUR_Yan.pdf.
Damian, Mihai. "Conception automatique de liaisons pivot par roulements utilisant les méthodes de l'intelligence artificielle." Toulouse 3, 1996. http://www.theses.fr/1996TOU30218.
Chorlay, Dominique. "Une Architecture d'environnement de résolution de problèmes en modélisation : Apports de l'intelligence artificielle." Compiègne, 1990. http://www.theses.fr/1990COMP0280.
Many kinds of knowledge are required to design an engineering software as well as to use it. The knowledge dealing with a particular software, is implicitly managed by both designers and users of this software. The one dealing with application fields is not always known by the user. Thus, we propose to integrate in an engineering software a knowledge base about its own working. This leads to a better communication between the user and its program as the software is able to answer requests and to manage automatically engineering tasks. Three moduli are needed to implement this intelligent dialogue. A natural language interface based on a semantic grammar which manages a specialist language as powerful and flexible as the language it is based on. A modulus, specific of the software, is able to create a sequence of commands to achieve a goal imposed by the user. Plan- generating systems has provided a good scheme for this modulus. An expert system is dealing with high level knowledge, i. E. The ones relevant from engineering application fields, but independent from the software itself. This modulus is also the manager which controles the way the system works
Liu, Zaisheng. "Application de l'intelligence artificielle a la caracterisation de l'etat des outils de coupe." Paris, ENSAM, 1988. http://www.theses.fr/1988ENAM0010.
SERVAJEAN, CHABARD MARIE-FRANCOISE. "Les apports de l'intelligence artificielle au genie logiciel : modelisation de l'activite de programmation." Clermont-Ferrand 2, 1990. http://www.theses.fr/1990CLF21292.
LEFEVRE, FREDERIC. "Application de l'intelligence artificielle a un systeme de conception d'exercices d'enseignement de comportement." Toulouse 3, 1988. http://www.theses.fr/1988TOU30012.
Rouxel, Marie. "Le refus de reconnaître le statut d'auteur à l'intelligence artificielle et ses conséquences." Master's thesis, Université Laval, 2019. http://hdl.handle.net/20.500.11794/38247.
« Artificial intelligence could be the most important event in the history of our civilization.” This sentence from Stephen Hawking demonstrates the importance that artificial intelligence can have in our society. The evolution of algorithms now allows artificial intelligence to create artistic creations in a virtually autonomous way since the user of artificial intelligence has only to start the process. This new way of creating raises many questions, including who the author of these creations is, and whether they can benefit from copyright protection. It appears that the criteria for granting copyright protection do not, in the current state of affairs, make it possible to integrate computer-generated creations with copyright. Indeed, these criteria do not allow to attribute the status of author to the artificial intelligence because AI is not a physical person. In the absence of protection by Copyright, one can wonder what happens to computer-generated creations. One of the consequences could be that the creations fall immediately into the public domain. Indeed, it would serve the public interest while keeping a very protective copyright of the person of the author.
CHERIE, NABIL. "Utilisation des techniques de l'intelligence artificielle pour la modelisation du mouvement d'objets animes." Paris 11, 1991. http://www.theses.fr/1991PA112075.
Fonkoua, Alain. "Une application de l'intelligence artificielle à la synthèse architecturale des circuits intégrés VLSI." Grenoble INPG, 1989. http://tel.archives-ouvertes.fr/tel-00335755.
Lefevre, Frédéric. "Application de l'intelligence artificielle à un système de conception d'exercices d'enseignement de comportements." Grenoble 2 : ANRT, 1988. http://catalogue.bnf.fr/ark:/12148/cb376150792.
Liu, Zaisheng. "Application de l'intelligence artificielle à la caractérisation de l'état des outils de coupe." Grenoble 2 : ANRT, 1988. http://catalogue.bnf.fr/ark:/12148/cb376153071.
Vidal, Jean-Philippe Dartus Denis Moisan Sabine Faure Jean-Baptiste. "Assistance au calage de modèles numériques en hydraulique fluviale apports de l'intelligence artificielle /." Toulouse : INP Toulouse, 2005. http://ethesis.inp-toulouse.fr/archive/00000116.
Fonkoua, Alain Blaise Mazaré Guy Mossière Jacques. "Une application de l'intelligence artificielle à la synthèse architecturale des circuits intégrés VLSI." S.l. : Université Grenoble 1, 2008. http://tel.archives-ouvertes.fr/tel-00335755.
Panaïotis, Thelma. "Distribution du plancton à diverses échelles : apport de l'intelligence artificielle pour l'écologie planctonique." Electronic Thesis or Diss., Sorbonne université, 2023. http://www.theses.fr/2023SORUS155.
As the basis of oceanic food webs and a key component of the biological carbon pump, planktonic organisms play major roles in the oceans. However, their small-scale distribution − governed by biotic interactions between organisms and interactions with the physico-chemical properties of the water masses in their immediate environment − are poorly described in situ due to the lack of suitable observation tools. New instruments performing high resolution imaging in situ in combination with machine learning algorithms to process the large amount of collected data now allows us to address these scales. The first part of this work focuses on the methodological development of two automated pipelines based on artificial intelligence. These pipelines allowed to efficiently detect planktonic organisms within raw images, and classify them into taxonomical or morphological categories. Then, in a second part, numerical ecology tools have been applied to study plankton distribution at different scales, using three different in situ imaging datasets. First, we investigated the link between plankton community and environmental conditions at the global scale. Then, we resolved plankton and particle distribution across a mesoscale front, and highlighted contrasted periods during the spring bloom. Finally, leveraging high frequency in situ imaging data, we investigated the fine-scale distribution and preferential position of Rhizaria, a group of understudied, fragile protists, some of which are mixotrophic. Overall, these studies demonstrate the effectiveness of in situ imaging combined with artificial intelligence to understand biophysical interactions in plankton and distribution patterns at small-scale
Clampitt, Megan. "Indexation de l'état de santé des coraux par une approche basée sur l'intelligence artificielle." Electronic Thesis or Diss., Université Côte d'Azur, 2023. http://www.theses.fr/2023COAZ6019.
Coral reefs are deteriorating at a startling rate and the development of fast and efficient monitoring schemas that attempt to evaluate coral health without only focusing on the absence or presence of disease or bleaching is essential. My Ph.D. research aims to combine the fields of Coral Biology, Computer Science, and Marine Conservation with the main question of my thesis being: how can artificial intelligence tools be used to assess coral health states from colony photographs? Since the assessment of individual coral colony health state remains poorly defined, our approach is to use AI tools to assess visual cues such as physically damaging conditions (boring organisms & predation), contact with other organisms (algae, sediment), and color changes that could correlate with health states. This was achieved by utilizing photographic data from the Tara Pacific Expedition to build the first version of AI machines capable of automatically recognizing these visual cues and then applying this tool to two types of field studies i). A longitudinal study set up in Moorea, French Polynesia aimed to investigate coral health as assessed by mortality/partial mortality events. ii). A comparative study between damaged, pristine, and restoration sites in Raja Ampat, Indonesia. The objective of these studies is to extract the visual cues that distinguish healthy from unhealthy corals. Thus, I was able to create an AI Model capable of automatically annotating coral colony photographs for visual cues relevant to the current health state of the colony
Bourgeade, Tom. "Interprétabilité a priori et explicabilité a posteriori dans le traitement automatique des langues." Thesis, Toulouse 3, 2022. http://www.theses.fr/2022TOU30063.
With the advent of Transformer architectures in Natural Language Processing a few years ago, we have observed unprecedented progress in various text classification or generation tasks. However, the explosion in the number of parameters, and the complexity of these state-of-the-art blackbox models, is making ever more apparent the now urgent need for transparency in machine learning approaches. The ability to explain, interpret, and understand algorithmic decisions will become paramount as computer models start becoming more and more present in our everyday lives. Using eXplainable AI (XAI) methods, we can for example diagnose dataset biases, spurious correlations which can ultimately taint the training process of models, leading them to learn undesirable shortcuts, which could lead to unfair, incomprehensible, or even risky algorithmic decisions. These failure modes of AI, may ultimately erode the trust humans may have otherwise placed in beneficial applications. In this work, we more specifically explore two major aspects of XAI, in the context of Natural Language Processing tasks and models: in the first part, we approach the subject of intrinsic interpretability, which encompasses all methods which are inherently easy to produce explanations for. In particular, we focus on word embedding representations, which are an essential component of practically all NLP architectures, allowing these mathematical models to process human language in a more semantically-rich way. Unfortunately, many of the models which generate these representations, produce them in a way which is not interpretable by humans. To address this problem, we experiment with the construction and usage of Interpretable Word Embedding models, which attempt to correct this issue, by using constraints which enforce interpretability on these representations. We then make use of these, in a simple but effective novel setup, to attempt to detect lexical correlations, spurious or otherwise, in some popular NLP datasets. In the second part, we explore post-hoc explainability methods, which can target already trained models, and attempt to extract various forms of explanations of their decisions. These can range from diagnosing which parts of an input were the most relevant to a particular decision, to generating adversarial examples, which are carefully crafted to help reveal weaknesses in a model. We explore a novel type of approach, in parts allowed by the highly-performant but opaque recent Transformer architectures: instead of using a separate method to produce explanations of a model's decisions, we design and fine-tune an architecture which jointly learns to both perform its task, while also producing free-form Natural Language Explanations of its own outputs. We evaluate our approach on a large-scale dataset annotated with human explanations, and qualitatively judge some of our approach's machine-generated explanations
Zarri, Gian Piero. "Utilisation de techniques relevant de l'intelligence artificielle pour le traitement de données biographiques complexes." Paris 11, 1985. http://www.theses.fr/1985PA112342.
The aim of this thesis is to provide a general description of RESEDA, an « intelligent » Information Retrieval system dealing with biographical data and using techniques borrowed from Knowledge Engineering and Artificial Intelligence (AI). All the system’s “knowledge” is represented in purely declarative form. This is the case both for the “fact database” and the “rule base”; the fact database contains the data, in the usual sens of the word that the system has to retrieve. Together, the fact and rule bases make up RESEDA’s “knowledge base”. Information in the knowledge base is depicted using a single knowledge representation language (“metalanguage”), which makes use of quantified variables when describing date in the rule base; the metalanguage is a particularly powerful realization of an AI type “case grammar”. For reasons of computational efficiency, the low-level (“level zero”) inferencing (retrieving) is carried out in RESEDA by using only the resources of the system’s match machine. This machine owes a large part of its power to the judicious use of temporal data in efficiently indexing the fact database. Only high-level inferences require the creation of real “inference engines”. RESEDA’s inference engine hat the general characteristics a) of being “event driven” in its initialization; b) of solving problems by constructing a “choice tree”. Traversal of the choice tree is performed depth-first with systematic backtracking. The high-level inference operations, relying on information in the rule base and making use of the inference engine, that are implemented in the system, are known as “transformations” and “hypotheses”. The “hypotheses” enable new causal relationships to be established between events in the fact database which are a priori totally disjointed; the system is thus equipped with an, albeit elementary, learning capability
Mathieu, Olivier. "Application des méthodes de l'intelligence artificielle à l'analyse de données en physique des particules." Aix-Marseille 2, 1990. http://www.theses.fr/1990AIX22058.
Yameogo, Relwende Aristide. "Risques et perspectives du big data et de l'intelligence artificielle : approche éthique et épistémologique." Thesis, Normandie, 2020. http://www.theses.fr/2020NORMLH10.
In the 21st century, the use of big data and AI in the field of health has gradually expanded, although it is accompanied by problems linked to the emergence of practices based on the use of digital traces. The aim of this thesis is to evaluate the use of big data and AI in medical practice, to discover the processes generated by digital tools in the field of health and to highlight the ethical problems they pose.The use of ICTs in medical practice is mainly based on the use of EHR, prescription software and connected objects. These uses raise many problems for physicians who are aware of the risk involved in protecting patients' health data. In this work, we are implementing a method for designing CDSS, the temporal fuzzy vector space. This method allows us to model a new clinical diagnostic score for pulmonary embolism. Through the "Human-trace" paradigm, our research allows us not only to measure the limitation in the use of ICT, but also to highlight the interpretative biases due to the delinking between the individual caught in his complexity as a "Human-trace" and the data circulating about him via digital traces. If big data, coupled with AI can play a major role in the implementation of CDSS, it cannot be limited to this field. We are also studying how to set up big data and AI development processes that respect the deontological and medical ethics rules associated with the appropriation of ICTs by the actors of the health system