Дисертації з теми "Self-supervised learning (artificial intelligence)"
Щоб переглянути інші типи публікацій з цієї теми, перейдіть за посиланням: Self-supervised learning (artificial intelligence).
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-50 дисертацій для дослідження на тему "Self-supervised learning (artificial intelligence)".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте дисертації для різних дисциплін та оформлюйте правильно вашу бібліографію.
1
Denize, Julien. "Self-supervised representation learning and applications to image and video analysis." Electronic Thesis or Diss., Normandie, 2023. http://www.theses.fr/2023NORMIR37.
Анотація:
Dans cette thèse, nous développons des approches d'apprentissage auto-supervisé pour l'analyse d'images et de vidéos. L'apprentissage de représentation auto-supervisé permet de pré-entraîner les réseaux neuronaux à apprendre des concepts généraux sans annotations avant de les spécialiser plus rapidement à effectuer des tâches, et avec peu d'annotations. Nous présentons trois contributions à l'apprentissage auto-supervisé de représentations d'images et de vidéos. Premièrement, nous introduisons le paradigme théorique de l'apprentissage contrastif doux et sa mise en œuvre pratique appelée Estimation Contrastive de Similarité (SCE) qui relie l'apprentissage contrastif et relationnel pour la représentation d'images. Ensuite, SCE est étendue à l'apprentissage de représentation vidéo temporelle globale. Enfin, nous proposons COMEDIAN, un pipeline pour l'apprentissage de représentation vidéo locale-temporelle pour l'architecture transformer. Ces contributions ont conduit à des résultats de pointe sur de nombreux benchmarks et ont donné lieu à de multiples contributions académiques et techniques publiéesIn this thesis, we develop approaches to perform self-supervised learning for image and video analysis. Self-supervised representation learning allows to pretrain neural networks to learn general concepts without labels before specializing in downstream tasks faster and with few annotations. We present three contributions to self-supervised image and video representation learning. First, we introduce the theoretical paradigm of soft contrastive learning and its practical implementation called Similarity Contrastive Estimation (SCE) connecting contrastive and relational learning for image representation. Second, SCE is extended to global temporal video representation learning. Lastly, we propose COMEDIAN a pipeline for local-temporal video representation learning for transformers. These contributions achieved state-of-the-art results on multiple benchmarks and led to several academic and technical published contributions
2
Nett, Ryan. "Dataset and Evaluation of Self-Supervised Learning for Panoramic Depth Estimation." DigitalCommons@CalPoly, 2020. https://digitalcommons.calpoly.edu/theses/2234.
Анотація:
Depth detection is a very common computer vision problem. It shows up primarily in robotics, automation, or 3D visualization domains, as it is essential for converting images to point clouds. One of the poster child applications is self driving cars. Currently, the best methods for depth detection are either very expensive, like LIDAR, or require precise calibration, like stereo cameras. These costs have given rise to attempts to detect depth from a monocular camera (a single camera). While this is possible, it is harder than LIDAR or stereo methods since depth can't be measured from monocular images, it has to be inferred. A good example is covering one eye: you still have some idea how far away things are, but it's not exact. Neural networks are a natural fit for this. Here, we build on previous neural network methods by applying a recent state of the art model to panoramic images in addition to pinhole ones and performing a comparative evaluation. First, we create a simulated depth detection dataset that lends itself to panoramic comparisons and contains pre-made cylindrical and spherical panoramas. We then modify monodepth2 to support cylindrical and cubemap panoramas, incorporating current best practices for depth detection on those panorama types, and evaluate its performance for each type of image using our dataset. We also consider the resources used in training and other qualitative factors.
3
Stanescu, Ana. "Semi-supervised learning for biological sequence classification." Diss., Kansas State University, 2015. http://hdl.handle.net/2097/35810.
Анотація:
Doctor of PhilosophyDepartment of Computing and Information Sciences
Doina Caragea
Successful advances in biochemical technologies have led to inexpensive, time-efficient production of massive volumes of data, DNA and protein sequences. As a result, numerous computational methods for genome annotation have emerged, including machine learning and statistical analysis approaches that practically and efficiently analyze and interpret data. Traditional machine learning approaches to genome annotation typically rely on large amounts of labeled data in order to build quality classifiers. The process of labeling data can be expensive and time consuming, as it requires domain knowledge and expert involvement. Semi-supervised learning approaches that can make use of unlabeled data, in addition to small amounts of labeled data, can help reduce the costs associated with labeling. In this context, we focus on semi-supervised learning approaches for biological sequence classification. Although an attractive concept, semi-supervised learning does not invariably work as intended. Since the assumptions made by learning algorithms cannot be easily verified without considerable domain knowledge or data exploration, semi-supervised learning is not always "safe" to use. Advantageous utilization of the unlabeled data is problem dependent, and more research is needed to identify algorithms that can be used to increase the effectiveness of semi-supervised learning, in general, and for bioinformatics problems, in particular. At a high level, we aim to identify semi-supervised algorithms and data representations that can be used to learn effective classifiers for genome annotation tasks such as cassette exon identification, splice site identification, and protein localization. In addition, one specific challenge that we address is the "data imbalance" problem, which is prevalent in many domains, including bioinformatics. The data imbalance phenomenon arises when one of the classes to be predicted is underrepresented in the data because instances belonging to that class are rare (noteworthy cases) or difficult to obtain. Ironically, minority classes are typically the most important to learn, because they may be associated with special cases, as in the case of splice site prediction. We propose two main techniques to deal with the data imbalance problem, namely a technique based on "dynamic balancing" (augmenting the originally labeled data only with positive instances during the semi-supervised iterations of the algorithms) and another technique based on ensemble approaches. The results show that with limited amounts of labeled data, semisupervised approaches can successfully leverage the unlabeled data, thereby surpassing their completely supervised counterparts. A type of semi-supervised learning, known as "transductive" learning aims to classify the unlabeled data without generalizing to new, previously not encountered instances. Theoretically, this aspect makes transductive learning particularly suitable for the task of genome annotation, in which an entirely sequenced genome is typically available, sometimes accompanied by limited annotation. We study and evaluate various transductive approaches (such as transductive support vector machines and graph based approaches) and sequence representations for the problems of cassette exon identification. The results obtained demonstrate the effectiveness of transductive algorithms in sequence annotation tasks.
4
Abou-Moustafa, Karim. "Metric learning revisited: new approaches for supervised and unsupervised metric learning with analysis and algorithms." Thesis, McGill University, 2012. http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=106370.
Анотація:
In machine learning one is usually given a data set of real high dimensional vectors X, based on which it is desired to select a hypothesis θ from the space of hypotheses Θ using a learning algorithm. An immediate assumption that is usually imposed on X is that it is a subset from the very general embedding space Rp which makes the Euclidean distance ∥•∥2 to become the default metric for the elements of X. Since various learning algorithms assume that the input space is Rp with its endowed metric ∥•∥2 as a (dis)similarity measure, it follows that selecting hypothesis θ becomes intrinsically tied to the Euclidean distance. Metric learning is the problem of selecting a specific metric dX from a certain family of metrics D based on the properties of the elements in the set X. Under some performance measure, the metric dX is expected to perform better on X than any other metric d 2 D. If the learning algorithm replaces the very general metric ∥•∥2 with the metric dX , then selecting hypothesis θ will be tied to the more specific metric dX which carries all the information on the properties of the elements in X. In this thesis I propose two algorithms for learning the metric dX ; the first for supervised learning settings, and the second for unsupervised, as well as for supervised and semi-supervised settings. In particular, I propose algorithms that take into consideration the structure and geometry of X on one hand, and the characteristics of real world data sets on the other. However, if we are also seeking dimensionality reduction, then under some mild assumptions on the topology of X, and based on the available a priori information, one can learn an embedding for X into a low dimensional Euclidean space Rp0, p0 << p, where the Euclidean distance better reveals the similarities between the elements of X and their groupings (clusters). That is, as a by-product, we obtain dimensionality reduction together with metric learning. In the supervised setting, I propose PARDA, or Pareto discriminant analysis for discriminative linear dimensionality reduction. PARDA is based on the machinery of multi-objective optimization; simultaneously optimizing multiple, possibly conflicting, objective functions. This allows PARDA to adapt to the class topology in the lower dimensional space, and naturally handles the class masking problem that is inherent in Fisher's discriminant analysis framework for multiclass problems. As a result, PARDA yields significantly better classification results when compared with modern techniques for discriminative dimensionality reduction. In the unsupervised setting, I propose an algorithmic framework, denoted by ?? (note the different notation), that encapsulates spectral manifold learning algorithms and gears them for metric learning. The framework ?? captures the local structure and the local density information from each point in a data set, and hence it carries all the information on the varying sample density in the input space. The structure of ?? induces two distance metrics for its elements, the Bhattacharyya-Riemann metric dBR and the Jeffreys-Riemann metric dJR. Both metrics reorganize the proximity between the points in X based on the local structure and density around each point. As a result, when combining the metric space (??, dBR) or (??, dJR) with spectral clustering and Euclidean embedding, they yield significant improvements in clustering accuracies and error rates for a large variety of clustering and classification tasks.Dans cette thèse, je propose deux algorithmes pour l'apprentissage de la métrique dX; le premier pour l'apprentissage supervisé, et le deuxième pour l'apprentissage non-supervisé, ainsi que pour l'apprentissage supervisé et semi-supervisé. En particulier, je propose des algorithmes qui prennent en considération la structure et la géométrie de X d'une part, et les caractéristiques des ensembles de données du monde réel d'autre part. Cependant, si on cherche également la réduction de dimension, donc sous certaines hypothèses légères sur la topologie de X, et en même temps basé sur des informations disponibles a priori, on peut apprendre une intégration de X dans un espace Euclidien de petite dimension Rp0 p0 << p, où la distance Euclidienne révèle mieux les ressemblances entre les éléments de X et leurs groupements (clusters). Alors, comme un sous-produit, on obtient simultanément une réduction de dimension et un apprentissage métrique. Pour l'apprentissage supervisé, je propose PARDA, ou Pareto discriminant analysis, pour la discriminante réduction linéaire de dimension. PARDA est basé sur le mécanisme d'optimisation à multi-objectifs; optimisant simultanément plusieurs fonctions objectives, éventuellement des fonctions contradictoires. Cela permet à PARDA de s'adapter à la topologie de classe dans un espace dimensionnel plus petit, et naturellement gère le problème de masquage de classe associé au discriminant Fisher dans le cadre d'analyse de problèmes à multi-classes. En conséquence, PARDA permet des meilleurs résultats de classification par rapport aux techniques modernes de réduction discriminante de dimension. Pour l'apprentissage non-supervisés, je propose un cadre algorithmique, noté par ??, qui encapsule les algorithmes spectraux d'apprentissage formant an algorithme d'apprentissage de métrique. Le cadre ?? capture la structure locale et la densité locale d'information de chaque point dans un ensemble de données, et donc il porte toutes les informations sur la densité d'échantillon différente dans l'espace d'entrée. La structure de ?? induit deux métriques de distance pour ses éléments: la métrique Bhattacharyya-Riemann dBR et la métrique Jeffreys-Riemann dJR. Les deux mesures réorganisent la proximité entre les points de X basé sur la structure locale et la densité autour de chaque point. En conséquence, lorsqu'on combine l'espace métrique (??, dBR) ou (??, dJR) avec les algorithmes de "spectral clustering" et "Euclidean embedding", ils donnent des améliorations significatives dans les précisions de regroupement et les taux d'erreur pour une grande variété de tâches de clustering et de classification.
5
Halpern, Yonatan. "Semi-Supervised Learning for Electronic Phenotyping in Support of Precision Medicine." Thesis, New York University, 2016. http://pqdtopen.proquest.com/#viewpdf?dispub=10192124.
Анотація:
Medical informatics plays an important role in precision medicine, delivering the right information to the right person, at the right time. With the introduction and widespread adoption of electronic medical records, in the United States and world-wide, there is now a tremendous amount of health data available for analysis.
Electronic record phenotyping refers to the task of determining, from an electronic medical record entry, a concise descriptor of the patient, comprising of their medical history, current problems, presentation, etc. In inferring such a phenotype descriptor from the record, a computer, in a sense, "understands'' the relevant parts of the record. These phenotypes can then be used in downstream applications such as cohort selection for retrospective studies, real-time clinical decision support, contextual displays, intelligent search, and precise alerting mechanisms.
We are faced with three main challenges:
First, the unstructured and incomplete nature of the data recorded in the electronic medical records requires special attention. Relevant information can be missing or written in an obscure way that the computer does not understand.
Second, the scale of the data makes it important to develop efficient methods at all steps of the machine learning pipeline, including data collection and labeling, model learning and inference.
Third, large parts of medicine are well understood by health professionals. How do we combine the expert knowledge of specialists with the statistical insights from the electronic medical record?
Probabilistic graphical models such as Bayesian networks provide a useful abstraction for quantifying uncertainty and describing complex dependencies in data. Although significant progress has been made over the last decade on approximate inference algorithms and structure learning from complete data, learning models with incomplete data remains one of machine learning’s most challenging problems. How can we model the effects of latent variables that are not directly observed?
The first part of the thesis presents two different structural conditions under which learning with latent variables is computationally tractable. The first is the "anchored'' condition, where every latent variable has at least one child that is not shared by any other parent. The second is the "singly-coupled'' condition, where every latent variable is connected to at least three children that satisfy conditional independence (possibly after transforming the data).
Variables that satisfy these conditions can be specified by an expert without requiring that the entire structure or its parameters be specified, allowing for effective use of human expertise and making room for statistical learning to do some of the heavy lifting. For both the anchored and singly-coupled conditions, practical algorithms are presented.
The second part of the thesis describes real-life applications using the anchored condition for electronic phenotyping. A human-in-the-loop learning system and a functioning emergency informatics system for real-time extraction of important clinical variables are described and evaluated.
The algorithms and discussion presented here were developed for the purpose of improving healthcare, but are much more widely applicable, dealing with the very basic questions of identifiability and learning models with latent variables - a problem that lies at the very heart of the natural and social sciences.
6
Taylor, Farrell R. "Evaluation of Supervised Machine Learning for Classifying Video Traffic." NSUWorks, 2016. http://nsuworks.nova.edu/gscis_etd/972.
Анотація:
Operational deployment of machine learning based classifiers in real-world networks has become an important area of research to support automated real-time quality of service decisions by Internet service providers (ISPs) and more generally, network administrators. As the Internet has evolved, multimedia applications, such as voice over Internet protocol (VoIP), gaming, and video streaming, have become commonplace. These traffic types are sensitive to network perturbations, e.g. jitter and delay. Automated quality of service (QoS) capabilities offer a degree of relief by prioritizing network traffic without human intervention; however, they rely on the integration of real-time traffic classification to identify applications. Accordingly, researchers have begun to explore various techniques to incorporate into real-world networks. One method that shows promise is the use of machine learning techniques trained on sub-flows – a small number of consecutive packets selected from different phases of the full application flow. Generally, research on machine learning classifiers was based on statistics derived from full traffic flows, which can limit their effectiveness (recall and precision) if partial data captures are encountered by the classifier. In real-world networks, partial data captures can be caused by unscheduled restarts/reboots of the classifier or data capture capabilities, network interruptions, or application errors. Research on the use of machine learning algorithms trained on sub-flows to classify VoIP and gaming traffic has shown promise, even when partial data captures are encountered. This research extends that work by applying machine learning algorithms trained on multiple sub-flows to classification of video streaming traffic.
Results from this research indicate that sub-flow classifiers have much higher and more consistent recall and precision than full flow classifiers when applied to video traffic. Moreover, the application of ensemble methods, specifically Bagging and adaptive boosting (AdaBoost) further improves recall and precision for sub-flow classifiers. Findings indicate sub-flow classifiers based on AdaBoost in combination with the C4.5 algorithm exhibited the best performance with the most consistent results for classification of video streaming traffic.
7
Coursey, Kino High. "An Approach Towards Self-Supervised Classification Using Cyc." Thesis, University of North Texas, 2006. https://digital.library.unt.edu/ark:/67531/metadc5470/.
Анотація:
Due to the long duration required to perform manual knowledge entry by human knowledge engineers it is desirable to find methods to automatically acquire knowledge about the world by accessing online information. In this work I examine using the Cyc ontology to guide the creation of Naïve Bayes classifiers to provide knowledge about items described in Wikipedia articles. Given an initial set of Wikipedia articles the system uses the ontology to create positive and negative training sets for the classifiers in each category. The order in which classifiers are generated and used to test articles is also guided by the ontology. The research conducted shows that a system can be created that utilizes statistical text classification methods to extract information from an ad-hoc generated information source like Wikipedia for use in a formal semantic ontology like Cyc. Benefits and limitations of the system are discussed along with future work.
8
Livi, Federico. "Supervised Learning with Graph Structured Data for Transprecision Computing." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/19714/.
Анотація:
Nell'era dell'Internet of things, dei Big Data e dell'industria 4.0, la crescente richiesta di risorse e strumenti atti ad elaborare la grande quantità di dati e di informazioni disponibili in ogni momento, ha posto l'attenzione su problemi oramai non più trascurabili inerenti al consumo di energia e ai costi che ne derivano. Si tratta del cosiddetto powerwall, ovvero della difficoltà fisica dei macchinari di sostenere il consumo di potenza necessario per il processamento di moli di dati sempre più grandi e per l'esecuzione di task sempre più sofisticati. Tra le nuove tecniche che si sono affermate negli ultimi anni per tentare di arginare questo problema è importante citare la cosiddetta Transprecision Computing, approccio che si impegna a migliorare il consumo dell'energia a discapito della precisione. Infatti, tramite la riduzione di bit di precisione nelle operazioni di floating point, è possibile ottenere una maggiore efficienza energetica ma anche una decrescita non lineare della precisione di computazione. A seconda del dominio di applicazione, questo tradeoff può portare effettivamente ad importanti miglioramenti, ma purtroppo risulta ancora complesso trovare la precisione ottimale per tutte le variabili rispettando nel mentre un limite superiore relativo all'errore. In letteratura, questo problema è perciò affrontato utilizzando euristiche e metodologie che coinvolgono direttamente modelli di ottimizzazione e di machine learning. Nel presente elaborato, si cerca di migliorare ulteriormente questi approcci, introducendo nuovi modelli di machine learning basati anche sull'analisi di relazioni complesse tra le variabili. In questo senso, si arriva anche ad esaminare tecniche che lavorano direttamente su dati strutturati a grafo, tramite lo studio di reti neurali più complesse, le cosiddette graph convolutional networks.
9
Rossi, Alex. "Self-supervised information retrieval: a novel approach based on Deep Metric Learning and Neural Language Models." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2021.
Анотація:
Most of the existing open-source search engines, utilize keyword or tf-idf based techniques to find relevant documents and web pages relative to an input query. Although these methods, with the help of a page rank or knowledge graphs, proved to be effective in some cases, they often fail to retrieve relevant instances for more complicated queries that would require a semantic understanding to be exploited.
In this Thesis, a self-supervised information retrieval system based on transformers is employed to build a semantic search engine over the library of Gruppo Maggioli company. Semantic search or search with meaning can refer to an understanding of the query, instead of simply finding words matches and, in general, it represents knowledge in a way suitable for retrieval.
We chose to investigate a new self-supervised strategy to handle the training of unlabeled data based on the creation of pairs of ’artificial’ queries and the respective positive passages. We claim that by removing the reliance on labeled data, we may use the large volume of unlabeled material on the web without being limited to languages or domains where labeled data is abundant.
10
Stroulia, Eleni. "Failure-driven learning as model-based self-redesign." Diss., Georgia Institute of Technology, 1994. http://hdl.handle.net/1853/8291.
11
Watkins, Andrew B. "AIRS: a resource limited artificial immune classifier." Master's thesis, Mississippi State : Mississippi State University, 2001. http://library.msstate.edu/etd/show.asp?etd=etd-11052001-102048.
12
Apprey-Hermann, Joseph Kwame. "Evaluating The Predictability of Pseudo-Random Number Generators Using Supervised Machine Learning Algorithms." Youngstown State University / OhioLINK, 2020. http://rave.ohiolink.edu/etdc/view?acc_num=ysu1588805461290138.
13
Nasrin, Mst Shamima. "Pathological Image Analysis with Supervised and Unsupervised Deep Learning Approaches." University of Dayton / OhioLINK, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=dayton1620052562772676.
14
Othmani-Guibourg, Mehdi. "Supervised learning for distribution of centralised multiagent patrolling strategies." Electronic Thesis or Diss., Sorbonne université, 2019. http://www.theses.fr/2019SORUS534.
Анотація:
Depuis presque deux décennies, la tâche de la patrouille a fait l'objet d'une attention toute particulière de la part de la communauté multi-agent. La patrouille multi-agent consiste à modéliser comme un système multi-agent une tâche de patrouille à optimiser. Cette optimisation revient à répartir dans l'espace et le temps les agents patrouilleurs sur la zone à surveiller, cela le plus efficacement possible; un tel problème constitue par là même un problème de décision. Un large éventail d'algorithmes basés sur des stratégies d’agent réactives, cognitives, d’apprentissage par renforcement, centralisées et décentralisées, entre autres, ont été développés pour rendre les stratégies de patrouille toujours plus performantes. Cependant, les approches existantes basées sur de l'apprentissage supervisé avaient peu été étudiées jusqu’à présent, bien que quelques travaux aient abordé cette question. L’idée principale et sous-jacente à l'apprentissage supervisé, qui n’est rien de plus qu’un ensemble de méthodes et d'outils permettant d’inférer de nouvelles connaissances, est d’apprendre une fonction associant à tout élément en entrée un élément en sortie, à partir d'un ensemble de données composé de paires d'éléments entrées-sorties; l'apprentissage, dans ce cas, permet au système de faire de la généralisation à de nouvelles données jamais observées auparavant. Jusqu'à présent, la meilleure stratégie de patrouille multi-agent en ligne, à savoir sans calcul préalable, s'est avérée être une stratégie centralisée à coordinateur. Cependant, comme pour tout processus de décision centralisé généralement, une telle stratégie est difficilement échelonnable. L'objectif de ce travail est alors de développer et de mettre en œuvre une nouvelle méthodologie visant à transformer toute stratégie centralisée performante en stratégie distribuée, c'est-à-dire par nature résiliente, plus adaptative aux changements de l'environnement et échelonnable. Ce faisant, le processus de décision centralisé, généralement représenté par un coordinateur dans la patrouille multi-agent, est distribué sur les agents patrouilleurs au moyen de méthodes d’apprentissage supervisé, de sorte que les agents de la stratégie distribuée résultante tendent chacun à capturer ou cristalliser une partie de l’algorithme exécuté par le processus de décision centralisé. Le résultat est alors un nouveau algorithme de prise de décision distribué, qui repose sur de l’apprentissage automatique. Dans cette thèse, une telle procédure de distribution de stratégie centralisée est établie, puis concrètement mise en œuvre en utilisant certaines architectures de réseaux de neurones. Ainsi, après avoir exposé le contexte et les motivations, nous posons la problématique étudiée. Les principales stratégies multi-agent élaborées jusqu'à présent dans le cadre de la patrouille multi-agent sont ensuite décrites, en particulier une stratégie centralisée à haute performance qui est la stratégie centralisée à distribuer ici étudiée, ainsi qu’une stratégie décentralisée assez simple qui est utilisée comme référence pour les stratégies décentralisées. Entre autres, quelques stratégies basées sur de l’apprentissage supervisé sont aussi décrites. Ensuite, le modèle ainsi que certains concept fondamentaux du problème de la patrouille multi-agent sont définisFor nearly two decades, patrolling has received significant attention from the multiagent community. Multiagent patrolling (MAP) consists in modelling a patrol task to optimise as a multiagent system. The problem of optimising a patrol task is to distribute the most efficiently agents over the area to patrol in space and time, which constitutes a decision-making problem. A range of algorithms based on reactive, cognitive, reinforcement learning, centralised and decentralised strategies, amongst others, have been developed to make such a task ever more efficient. However, the existing patrolling-specific approaches based on supervised learning were still at preliminary stages, although a few works addressed this issue. Central to supervised learning, which is a set of methods and tools that allow inferring new knowledge, is the idea of learning a function mapping any input to an output from a sample of data composed of input-output pairs; learning, in this case, enables the system to generalise to new data never observed before. Until now, the best online MAP strategy, namely without precalculation, has turned out to be a centralised strategy with a coordinator. However, as for any centralised decision process in general, such a strategy is hardly scalable. The purpose of this work is then to develop and implement a new methodology aiming at turning any high-performance centralised strategy into a distributed strategy. Indeed, distributed strategies are by design resilient, more adaptive to changes in the environment, and scalable. In doing so, the centralised decision process, generally represented in MAP by a coordinator, is distributed into patrolling agents by means of supervised learning methods, so that each agent of the resultant distributed strategy tends to capture a part of the algorithm executed by the centralised decision process. The outcome is a new distributed decision-making algorithm based on machine learning. In this dissertation therefore, such a procedure of distribution of centralised strategy is established, then concretely implemented using some artificial neural networks architectures. By doing so, after having exposed the context and motivations of this work, we pose the problematic that led our study. The main multiagent strategies devised until now as part of MAP are then described, particularly a high-performance coordinated strategy, which is the centralised strategy studied in this work, as well as a simple decentralised strategy used as reference for decentralised strategies. Among others, some existing strategies based on supervised learning are also described. Thereafter, the model as well as certain of key concepts of MAP are defined. We also define the methodology laid down to address and study this problematic. This methodology comes in the form of a procedure that allows decentralising any centralised strategy by means of supervised learning. Then, the software ecosystem we developed for the needs of this work is also described, particularly PyTrol a discrete-time simulator dedicated to MAP developed with the aim of performing MAP simulation, to assess strategies and generate data, and MAPTrainer, a framework hinging on the PyTorch machine learning library, dedicated to research in machine learning in the context of MAP
15
Mao, Yi. "Domain knowledge, uncertainty, and parameter constraints." Diss., Georgia Institute of Technology, 2010. http://hdl.handle.net/1853/37295.
16
Calderon-Vilca, Hugo D., William I. Wun-Rafael, and Roberto Miranda-Loarte. "Simulation of suicide tendency by using machine learning." IEEE Computer Society, 2018. http://hdl.handle.net/10757/624720.
Анотація:
El texto completo de este trabajo no está disponible en el Repositorio Académico UPC por restricciones de la casa editorial donde ha sido publicado.Suicide is one of the most distinguished causes of death on the news worldwide. There are several factors and variables that can lead a person to commit this act, for example, stress, self-esteem, depression, among others. The causes and profiles of suicide cases are not revealed in detail by the competent institutions. We propose a simulation with a systematically generated dataset; such data reflect the adolescent population with suicidal tendency in Peru. We will evaluate three algorithms of supervised machine learning as a result of the algorithm C4.5 which is based on the trees to classify in a better way the suicidal tendency of adolescents. We finally propose a desktop tool that determines the suicidal tendency level of the adolescent.
Revisión por pares
17
Charnay, Clément. "Enhancing supervised learning with complex aggregate features and context sensitivity." Thesis, Strasbourg, 2016. http://www.theses.fr/2016STRAD025/document.
Анотація:
Dans cette thèse, nous étudions l'adaptation de modèles en apprentissage supervisé. Nous adaptons des algorithmes d'apprentissage existants à une représentation relationnelle. Puis, nous adaptons des modèles de prédiction aux changements de contexte.En représentation relationnelle, les données sont modélisées par plusieurs entités liées par des relations. Nous tirons parti de ces relations avec des agrégats complexes. Nous proposons des heuristiques d'optimisation stochastique pour inclure des agrégats complexes dans des arbres de décisions relationnels et des forêts, et les évaluons sur des jeux de données réelles.Nous adaptons des modèles de prédiction à deux types de changements de contexte. Nous proposons une optimisation de seuils sur des modèles à scores pour s'adapter à un changement de coûts. Puis, nous utilisons des transformations affines pour adapter les attributs numériques à un changement de distribution. Enfin, nous étendons ces transformations aux agrégats complexesIn this thesis, we study model adaptation in supervised learning. Firstly, we adapt existing learning algorithms to the relational representation of data. Secondly, we adapt learned prediction models to context change.In the relational setting, data is modeled by multiples entities linked with relationships. We handle these relationships using complex aggregate features. We propose stochastic optimization heuristics to include complex aggregates in relational decision trees and Random Forests, and assess their predictive performance on real-world datasets.We adapt prediction models to two kinds of context change. Firstly, we propose an algorithm to tune thresholds on pairwise scoring models to adapt to a change of misclassification costs. Secondly, we reframe numerical attributes with affine transformations to adapt to a change of attribute distribution between a learning and a deployment context. Finally, we extend these transformations to complex aggregates
18
Buchi, Baptiste. "Learning system for self-reconfiguration of micro-robot networks." Electronic Thesis or Diss., Bourgogne Franche-Comté, 2023. http://www.theses.fr/2023UBFCA017.
Анотація:
Le problème d'auto-reconfiguration des réseaux de micro-robots est l'un des défis majeurs de la robotique modulaire. Un ensemble de micro-robots reliés par des liens électromagnétiques ou mécaniques se réorganisent afin d'atteindre des formes cibles données. Le problème d'auto-reconfiguration est un problème complexe pour trois raisons. Premièrement, le nombre de configurations distinctes d'un réseau de robots modulaires est très élevé. Deuxièmement, comme les modules sont libres de se mouvoir indépendamment les uns des autres, à partir de chaque configuration il est possible d'atteindre un nombre d'autres configurations lui aussi très élevé. Troisièmement et comme conséquence du précédent point, l'espace de recherche reliant deux configurations est exponentiel ce qui empêche la détermination du planning optimal de l'auto-reconfiguration.Nous proposons dans ce travail, dans un premier temps, une approche d'auto-reconfiguration autonome distribuée TBSR, axée sur l'optimisation des déplacements pour une meilleure répartition des tâches. En d'autres termes, il s'agit de répartir l'effort fourni par chaque robot pour atteindre la forme finale.Dans un deuxième temps, nous proposons des approches hybrides qui tirent profit des avantages des méthodes centralisées et des méthodes distribuées. Ces approches permettent de sélectionner le meilleur algorithme distribué avant le lancement de la procédure de reconfiguration. Une gamme d'algorithmes distribués sont préalablement installés sur chaque robot modulaire. Au début de la procédure d'auto-reconfiguration, un coordinateur diffuse à l'ensemble des micro-robots, les données relatives à la forme finale à atteindre et l'algorithme distribué.Pour ce faire, nous avons déterminé les caractéristiques pertinentes des problèmes d'auto-reconfiguration permettant d'identifier l'approche algorithmique la plus adaptée.Une étude de l'impact de chaque méthode de reconfiguration et des paramètres de performances a été menée pour établir une base de connaissances. Cette base consigne les performances des divers algorithmes en fonction de différents paramètres pour un éventail varié de scénarios de problèmes d'auto-reconfiguration.A l'aide d'un système de classification, il est ainsi possible d'établir pour chaque méthode d'auto-reconfiguration les caractéristiques des scénarios d'auto-reconfiguration pour lesquels elle se montre efficace. Les mécanismes d'apprentissage développés IA (e.g., réseaux de neurones) sont mis en œuvre. Une première approche hybride CNNSR proposée fait appel aux réseaux de neurones artificiels pour prédire l’approche optimale pour l'auto-reconfiguration. Une approche CNN2SR (une version améliorée de CNNSR), a été introduite pour la précision et la réduction des erreurs, en affinant la classification.Dans un troisième temps, une modélisation de la consommation énergétique, issue d'expérimentations réelles avec des robots modulaires physiques (Catom 2D) a été établie. Cela a permis de mettre en œuvre une troisième approche hybride CNN3SR axé sur l'optimisation énergétique pour les robots modulairesThe problem of self-reconfiguration of micro-robot networks is one of the major challenges of modular robotics. A set of micro-robots connected by electromagnetic or mechanical links reorganize themselves in order to reach given target shapes. The self-reconfiguration problem is a complex problem for three reasons. First, the number of distinct configurations of a modular robot network is very high. Secondly, as the modules are free to move independently of each other, from each configuration it is possible to reach a very high number of other configurations. Thirdly and as a consequence of the previous point, the search space connecting two configurations is exponential which prevents the determination of the optimal schedule of the self-reconfiguration.In this work, we propose, firstly, a distributed autonomous self-reconfiguration approach TBSR, focused on the optimization of movements for a better distribution of tasks. In other words, it involves distributing the effort made by each robot to reach the final shape.Secondly, we propose hybrid approaches that take advantage of the advantages of centralized methods and distributed methods. These approaches make it possible to select the best distributed algorithm before launching the reconfiguration procedure. A range of distributed algorithms are pre-installed on each modular robot. At the start of the self-reconfiguration procedure, a coordinator broadcasts to all the micro-robots the data relating to the final shape to be achieved and the distributed algorithm.To do this, we determined the relevant characteristics of self-reconfiguration problems allowing us to identify the most suitable algorithmic approach.A study of the impact of each reconfiguration method and performance parameters was conducted to establish a knowledge base. This database records the performance of various algorithms based on different parameters for a diverse range of self-reconfiguration problem scenarios.Using a classification system, it is thus possible to establish for each self-reconfiguration method the characteristics of the self-reconfiguration scenarios for which it is effective. The learning mechanisms developed by AI (e.g., neural networks) are implemented. A first proposed hybrid CNNSR approach uses artificial neural networks to predict the optimal approach for self-reconfiguration. A CNN2SR approach (an improved version of CNNSR), was introduced for accuracy and error reduction, by refining the classification.Thirdly, a modeling of energy consumption, resulting from real experiments with physical modular robots (Catom 2D) was established. This made it possible to implement a third hybrid CNN3SR approach focused on energy optimization for modular robots
19
Dhyani, Dushyanta Dhyani. "Boosting Supervised Neural Relation Extraction with Distant Supervision." The Ohio State University, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=osu1524095334803486.
20
Balasubramanian, Krishnakumar. "Learning without labels and nonnegative tensor factorization." Thesis, Georgia Institute of Technology, 2010. http://hdl.handle.net/1853/33926.
Анотація:
Supervised learning tasks like building a classifier, estimating the error rate of the
predictors, are typically performed with labeled data. In most cases, obtaining labeled data
is costly as it requires manual labeling. On the other hand, unlabeled data is available in
abundance. In this thesis, we discuss methods to perform supervised learning tasks with
no labeled data. We prove consistency of the proposed methods and demonstrate its applicability
with synthetic and real world experiments. In some cases, small quantities of labeled data maybe easily available and supplemented with large quantities of unlabeled data (semi-supervised learning). We derive the asymptotic efficiency of generative models for semi-supervised learning and quantify the effect of labeled and unlabeled data on the quality of the estimate. Another independent track of the thesis is efficient computational methods for nonnegative tensor factorization (NTF). NTF provides the user with rich modeling capabilities but it comes with an added computational cost. We provide a fast algorithm for performing NTF using a modified active set method called block principle pivoting method and demonstrate its applicability to social network analysis and text
mining.
21
Minakshi, Mona. "A Machine Learning Framework to Classify Mosquito Species from Smart-phone Images." Scholar Commons, 2018. https://scholarcommons.usf.edu/etd/7340.
Анотація:
Mosquito borne diseases have been a constant scourge across the globe resulting in numerous diseases with debilitating consequences, and also death. To derive trends on population of mosquitoes in an area, trained personnel lay traps, and after collecting trapped specimens, they spend hours under a microscope to inspect each specimen for identifying the actual species and logging it. This is vital, because multiple species of mosquitoes can reside in any area, and the vectors that some of them carry are not the same ones carried by others. The species identification process is naturally laborious, and imposes severe cognitive burden, since sometimes, hundreds of mosquitoes can get trapped. Most importantly, common citizens cannot aid in this task. In this paper, we design a system based on smart-phone images for mosquito species identification, that integrates image processing, feature selection, unsupervised clustering, and support vector machine based algorithm for classification. Results with a total of 101 female mosquito specimens spread across 9 different vector carrying species (that were captured from a real outdoor trap) demonstrate an overall accuracy of 77% in species identification. When implemented as a smart-phone app, the latency and energy consumption were minimal. In terms of practical impact, common citizens can benefit from our system to identify mosquito species by themselves, and also share images to local/ global mosquito control centers. In economically disadvantaged areas across the globe, tools like these can enable novel citizen-science enabled mechanisms to combat spread of mosquitoes.
22
Yu, Chen-Ping. "Computational model of MST neuron receptive field and interaction effect for the perception of self-motion /." Online version of thesis, 2008. http://hdl.handle.net/1850/9588.
23
Campbell, Benjamin W. "Supervised and Unsupervised Machine Learning Strategies for Modeling Military Alliances." The Ohio State University, 2019. http://rave.ohiolink.edu/etdc/view?acc_num=osu1558024695617708.
24
Kilinc, Ismail Ozsel. "Graph-based Latent Embedding, Annotation and Representation Learning in Neural Networks for Semi-supervised and Unsupervised Settings." Scholar Commons, 2017. https://scholarcommons.usf.edu/etd/7415.
Анотація:
Machine learning has been immensely successful in supervised learning with outstanding examples in major industrial applications such as voice and image recognition. Following these developments, the most recent research has now begun to focus primarily on algorithms which can exploit very large sets of unlabeled examples to reduce the amount of manually labeled data required for existing models to perform well. In this dissertation, we propose graph-based latent embedding/annotation/representation learning techniques in neural networks tailored for semi-supervised and unsupervised learning problems. Specifically, we propose a novel regularization technique called Graph-based Activity Regularization (GAR) and a novel output layer modification called Auto-clustering Output Layer (ACOL) which can be used separately or collaboratively to develop scalable and efficient learning frameworks for semi-supervised and unsupervised settings.
First, singularly using the GAR technique, we develop a framework providing an effective and scalable graph-based solution for semi-supervised settings in which there exists a large number of observations but a small subset with ground-truth labels. The proposed approach is natural for the classification framework on neural networks as it requires no additional task calculating the reconstruction error (as in autoencoder based methods) or implementing zero-sum game mechanism (as in adversarial training based methods). We demonstrate that GAR effectively and accurately propagates the available labels to unlabeled examples. Our results show comparable performance with state-of-the-art generative approaches for this setting using an easier-to-train framework.
Second, we explore a different type of semi-supervised setting where a coarse level of labeling is available for all the observations but the model has to learn a fine, deeper level of latent annotations for each one. Problems in this setting are likely to be encountered in many domains such as text categorization, protein function prediction, image classification as well as in exploratory scientific studies such as medical and genomics research. We consider this setting as simultaneously performed supervised classification (per the available coarse labels) and unsupervised clustering (within each one of the coarse labels) and propose a novel framework combining GAR with ACOL, which enables the network to perform concurrent classification and clustering. We demonstrate how the coarse label supervision impacts performance and the classification task actually helps propagate useful clustering information between sub-classes. Comparative tests on the most popular image datasets rigorously demonstrate the effectiveness and competitiveness of the proposed approach.
The third and final setup builds on the prior framework to unlock fully unsupervised learning where we propose to substitute real, yet unavailable, parent- class information with pseudo class labels. In this novel unsupervised clustering approach the network can exploit hidden information indirectly introduced through a pseudo classification objective. We train an ACOL network through this pseudo supervision together with unsupervised objective based on GAR and ultimately obtain a k-means friendly latent representation. Furthermore, we demonstrate how the chosen transformation type impacts performance and helps propagate the latent information that is useful in revealing unknown clusters. Our results show state-of-the-art performance for unsupervised clustering tasks on MNIST, SVHN and USPS datasets with the highest accuracies reported to date in the literature.
25
Leoni, Cristian. "Interpretation of Dimensionality Reduction with Supervised Proxies of User-defined Labels." Thesis, Linnéuniversitetet, Institutionen för datavetenskap och medieteknik (DM), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:lnu:diva-105622.
Анотація:
Research on Machine learning (ML) explainability has received a lot of focus in recent times. The interest, however, mostly focused on supervised models, while other ML fields have not had the same level of attention. Despite its usefulness in a variety of different fields, unsupervised learning explainability is still an open issue. In this paper, we present a Visual Analytics framework based on eXplainable AI (XAI) methods to support the interpretation of Dimensionality reduction methods. The framework provides the user with an interactive and iterative process to investigate and explain user-perceived patterns for a variety of DR methods by using XAI methods to explain a supervised method trained on the selected data. To evaluate the effectiveness of the proposed solution, we focus on two main aspects: the quality of the visualization and the quality of the explanation. This challenge is tackled using both quantitative and qualitative methods, and due to the lack of pre-existing test data, a new benchmark has been created. The quality of the visualization is established using a well-known survey-based methodology, while the quality of the explanation is evaluated using both case studies and a controlled experiment, where the generated explanation accuracy is evaluated on the proposed benchmark. The results show a strong capacity of our framework to generate accurate explanations, with an accuracy of 89% over the controlled experiment. The explanation generated for the two case studies yielded very similar results when compared with pre-existing, well-known literature on ground truths. Finally, the user experiment generated high quality overall scores for all assessed aspects of the visualization.
26
Cao, Xi Hang. "On Leveraging Representation Learning Techniques for Data Analytics in Biomedical Informatics." Diss., Temple University Libraries, 2019. http://cdm16002.contentdm.oclc.org/cdm/ref/collection/p245801coll10/id/586006.
Анотація:
Computer and Information SciencePh.D.
Representation Learning is ubiquitous in state-of-the-art machine learning workflow, including data exploration/visualization, data preprocessing, data model learning, and model interpretations. However, the majority of the newly proposed Representation Learning methods are more suitable for problems with a large amount of data. Applying these methods to problems with a limited amount of data may lead to unsatisfactory performance. Therefore, there is a need for developing Representation Learning methods which are tailored for problems with ``small data", such as, clinical and biomedical data analytics. In this dissertation, we describe our studies of tackling the challenging clinical and biomedical data analytics problem from four perspectives: data preprocessing, temporal data representation learning, output representation learning, and joint input-output representation learning. Data scaling is an important component in data preprocessing. The objective in data scaling is to scale/transform the raw features into reasonable ranges such that each feature of an instance will be equally exploited by the machine learning model. For example, in a credit flaw detection task, a machine learning model may utilize a person's credit score and annual income as features, but because the ranges of these two features are different, a machine learning model may consider one more heavily than another. In this dissertation, I thoroughly introduce the problem in data scaling and describe an approach for data scaling which can intrinsically handle the outlier problem and lead to better model prediction performance. Learning new representations for data in the unstandardized form is a common task in data analytics and data science applications. Usually, data come in a tubular form, namely, the data is represented by a table in which each row is a feature (row) vector of an instance. However, it is also common that the data are not in this form; for example, texts, images, and video/audio records. In this dissertation, I describe the challenge of analyzing imperfect multivariate time series data in healthcare and biomedical research and show that the proposed method can learn a powerful representation to encounter various imperfections and lead to an improvement of prediction performance. Learning output representations is a new aspect of Representation Learning, and its applications have shown promising results in complex tasks, including computer vision and recommendation systems. The main objective of an output representation algorithm is to explore the relationship among the target variables, such that a prediction model can efficiently exploit the similarities and potentially improve prediction performance. In this dissertation, I describe a learning framework which incorporates output representation learning to time-to-event estimation. Particularly, the approach learns the model parameters and time vectors simultaneously. Experimental results do not only show the effectiveness of this approach but also show the interpretability of this approach from the visualizations of the time vectors in 2-D space. Learning the input (feature) representation, output representation, and predictive modeling are closely related to each other. Therefore, it is a very natural extension of the state-of-the-art by considering them together in a joint framework. In this dissertation, I describe a large-margin ranking-based learning framework for time-to-event estimation with joint input embedding learning, output embedding learning, and model parameter learning. In the framework, I cast the functional learning problem to a kernel learning problem, and by adopting the theories in Multiple Kernel Learning, I propose an efficient optimization algorithm. Empirical results also show its effectiveness on several benchmark datasets.
Temple University--Theses
27
Budnyk, Ivan. "Contribution to the Study and Implementation of Intelligent Modular Self-organizing Systems." Phd thesis, Université Paris-Est, 2009. http://tel.archives-ouvertes.fr/tel-00481367.
Анотація:
Les problèmes de la classification ont reçu une attention considérable dans des différents champs d'ingénierie comme traitement des images biomédicales, identification a partir de la voix, reconnaissance d'empreinte digitale etc. Les techniques d'intelligence artificielles, incluant les réseaux de neurones artificiels, permettent de traiter des problèmes de ce type. En particulier, les problèmes rencontrés nécessitent la manipulation de bases de données de tailles très importantes. Des structures de traitement adaptatives et exploitant des ensembles de classificateurs sont utilisées. Dans cette thèse, nous décrivons principalement le développement et des améliorations apportées à un outil de classification désigné par le terme Tree-like Divide to Simplify ou T-DTS. Nos efforts se sont portés sur l'un des modules de cet outil, le module d'estimation de complexité. L'architecture de l'outil T-DTS est très flexible et nécessite le choix d'un nombre important de paramètres. Afin de simplifier l'exploitation de T-DTS, nous avons conçu et développé une procédure automatique d'optimisation d'un de ces plus importants paramètres, le seuil de décision associé à la mesure de complexité. La contribution principale de cette thèse concerne le développement de modules pouvant s'implanté sur une architecture de calcul matérielle parallèle. Ce ceci permet de se rapproché d'une implantation purement matérielle de l'outil T-DTS
28
Kim, Seungyeon. "Novel document representations based on labels and sequential information." Diss., Georgia Institute of Technology, 2015. http://hdl.handle.net/1853/53946.
Анотація:
A wide variety of text analysis applications are based on statistical machine learning techniques. The success of those applications is critically affected by how we represent a document. Learning an efficient document representation has two major challenges: sparsity and sequentiality. The sparsity often causes high estimation error, and text's sequential nature, interdependency between words, causes even more complication.
This thesis presents novel document representations to overcome the two challenges. First, I employ label characteristics to estimate a compact document representation. Because label attributes implicitly describe the geometry of dense subspace that has substantial impact, I can effectively resolve the sparsity issue while only focusing the compact subspace. Second, while modeling a document as a joint or conditional distribution between words and their sequential information, I can efficiently reflect sequential nature of text in my document representations. Lastly, the thesis is concluded with a document representation that employs both labels and sequential information in a unified formulation.
The following four criteria are utilized to evaluate the goodness of representations: how close a representation is to its original data, how strongly a representation can be distinguished from each other, how easy to interpret a representation by a human, and how much computational effort is needed for a representation.
While pursuing those good representation criteria, I was able to obtain document representations that are closer to the original data, stronger in discrimination, and easier to be understood than traditional document representations. Efficient computation algorithms make the proposed approaches largely scalable. This thesis examines emotion prediction, temporal emotion analysis, modeling documents with edit histories, locally coherent topic modeling, and text categorization tasks for possible applications.
29
Velander, Alice, and Harrysson David Gumpert. "Do Judge a Book by its Cover! : Predicting the genre of book covers using supervised deep learning. Analyzing the model predictions using explanatory artificial intelligence methods and techniques." Thesis, Linköpings universitet, Datorseende, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-177691.
Анотація:
In Storytel’s application on which a user can read and listen to digitalized literature, a user is displayed a list of books where the first thing the user encounters is the book title and cover. A book cover is therefore essential to attract a consumer’s attention. In this study, we take a data-driven approach to investigate the design principles for book covers through deep learning models and explainable AI. The first aim is to explore how well a Convolutional Neural Network (CNN) can interpret and classify a book cover image according to its genre in a multi-class classification task. The second aim is to increase model interpretability and investigate model feature to genre correlations. With the help of the explanatory artificial intelligence method Gradient-weighted Class Activation Map (Grad-CAM), we analyze the pixel-wise contribution to the model prediction. In addition, object detection by YOLOv3 was implemented to investigate which objects are detectable and reoccurring in the book covers. An interplay between Grad-CAM and YOLOv3 was used to investigate how identified objects and features correlate to a specific book genre and ultimately answer what makes a good book cover. Using a State-of-the-Art CNN model architecture we achieve an accuracy of 48% with the best class-wise accuracies for genres Erotica, Economy & Business and Children with accuracies 73%, 67% and 66%. Quantitative results from the Grad-CAM and YOLOv3 interplay show some strong associations between objects and genres, while indicating weak associations between abstract design principles and genres. Furthermore, a qualitative analysis of Grad-CAM visualizations show strong relevance of certain objects and text fonts for specific book genres. It was also observed that the portrayal of a feature was relevant for the model prediction of certain genres.
30
Lindkvist, Emilie. "Learning-by-modeling : Novel Computational Approaches for Exploring the Dynamics of Learning and Self-governance in Social-ecological Systems." Doctoral thesis, Stockholms universitet, Stockholm Resilience Centre, 2016. http://urn.kb.se/resolve?urn=urn:nbn:se:su:diva-122395.
Анотація:
As a consequence of global environmental change, sustainable management and governance of natural resources face critical challenges, such as dealing with non-linear dynamics, increased resource variability, and uncertainty. This thesis seeks to address some of these challenges by using simulation models. The first line of research focuses on the use of learning-by-doing (LBD) for managing a renewable resource, exemplified by a fish stock, and explores LBD in a theoretical model using artificial intelligence (Paper I and II). The second line of research focuses on the emergence of different forms of self-governance and their interrelation with the dynamics of trust among fishers when harvesting a shared resource, using an agent-based model. This model is informed by qualitative data based on small-scale fisheries in Mexico (Paper III and IV). Paper I and II find that the most sustainable harvesting strategy requires that the actor values current and future yields equally, cautiously experiments around what is perceived as the best harvest action, and rapidly updates its ‘mental model’ to any perceived change in catch. More specifically, Paper II reveals that understanding these aspects in relation to the type of change can yield not only increased performance, but also, and more importantly, increased robustness to both fast and slow changes in resource dynamics. However, when resource dynamics include the possibility of a more fundamental shift in system characteristics (a regime shift), LBD is problematic due to the potential for crossing a threshold, resulting in possible persistent reductions in harvests (Paper I). In Paper III, results indicate that cooperative forms of self-governance are more likely to establish and persist in communities where fishers’ have prior cooperative experience, fishers’ trustworthiness is more or less equal, and that this likelihood increases when resource availability fluctuates seasonally. Finally, to achieve a transformation toward more cooperative forms of self-governance, interventions are required that can strengthen both financial capital and trust among the members of the cooperatives (Paper IV). The unique contribution of this thesis lies in the method for ‘quantitatively’ studying LBD, the stylized model of a small-scale fishery, and the analysis of the two models to advance our understanding of processes of learning and self-governance in uncertain and variable social-ecological environments. Together, the results shed light on how social and ecological factors and processes co-evolve to shape social-ecological outcomes, as well as contributing to the development of novel methods within the emerging field of sustainability science.I vårt antropocena tidevarv är ett långsiktigt förvaltarskap av naturresurser inom social-ekologiska system av yttersta vikt. Detta kräver en djup förståelse av människan, ekologin, interaktionerna sinsemellan och deras utveckling över tid. Syftet med denna avhandling är att nå en djupare och mer nyanserad förståelse kring två av grundpelarna inom forskningen av hållbar förvaltning av naturresurser–kontinuerligt lärande genom learning-by-doing (LBD) för att förstå naturresursens dynamik, samt vad som kan kallas socialt kapital, i detta sammanhang i betydelsen tillit mellan individer, som naturligtvis ligger till grund för framgångsrik gemensam förvaltning. Denna föresats operationaliseras genom att använda två olika simuleringsmodeller. Den ena modellen undersöker hur en hållbar förvaltning av en förnyelsebar resurs, i denna avhandling exemplifierad av en fiskepopulation, kan uppnås genom LBD. Den andra modellen söker blottlägga det komplexa sociala samspel som krävs för att praktisera gemensam förvaltning genom att använda ett fiskesamhälle som fallstudie. Tidigare forskning på båda dessa två områden är relativt omfattade. Emellertid har den forskning som specialiserat sig på LBD i huvudsak inskränkt sig till empiriska fallstudier. Vad som bryter ny mark i denna avhandling är att vi konstruerar en simuleringsmodell av LBD där vi kan studera lärandeprocessen i detalj för att uppnå en mer hållbar förvaltning över tid. Beträffande modellen som behandlar socialt kapital så har tidigare forskning fokuserat på hur en organisation, eller grupp, kan uppnå hållbar förvaltning. Dock saknas ett helhetsgrepp där som tar hänsyn till alla nivåer; från individnivå (mikro), via gruppnivå (meso), till samhällsnivå (makro). Detta är något som denna avhandling försöker avhjälpa genom att undersöka betydelsen av individers egenskaper, uppbyggnaden av socialt kapital, samt hur detta påverkar emergens av ett samhälle dominerat av mer kooperativa förvaltningsformer respektive mer hierarkiska diton. I papper I and II studeras kärnan av LBD som återkoppling mellan en aktör och en resurs, där aktören lär sig genom upprepade interaktioner med en resurs. Resultaten visar att LBD är av avgörande betydelse för en hållbar förvaltning, speciellt då naturresursens dynamik är stadd i förändring. I den mest hållbara strategin bör aktören värdera nuvarande och framtida fångster lika högt, försiktigt experimentera kring vad aktören upplever som bästa strategi, för att sedan anpassa sin mentala modell till upplevda förändringar i fångst relativt dess insats någorlunda kraftigt. I papper III och IV behandlas uppbyggnaden av förtroende mellan individer och grupp, samt själv-organiserat styre. Genom att använda småskaligt fiske i Mexiko som en illustrativ fallstudie, utvecklades en agent-baserad modell av ett arketypiskt småskaligt fiskesamhälle. Resultaten indikerar att kooperativa förvaltningsformer är mer dominanta i samhällen där de som utför fisket har liknande pålitlighet, starkt gemensamt socialt kapital vid kooperativets start, och då resursen fluktuerar säsongsmässigt (papper III). Papper IV visar att för att uppnå en transformation från hierarkiska förvaltningsformer till kooperativa diton krävs interventioner som inriktar sig på både socialt och finansiellt kapital. Denna avhandling bidrar således till en djupare förståelse kring hur socialt kapital växer fram, samt hur mer strategiska LBD processer bör utformas när abrupta och osäkra förändringar i ekosystemen blir allt vanligare på grund av människans ökade tryck på planeten.
At the time of the doctoral defense, the following papers were unpublished and had a status as follows: Paper 2: Submitted. Paper 3: Submitted. Paper 4: Manuscript.
31
Jezequel, Loïc. "Vers une détection d'anomalie unifiée avec une application à la détection de fraude." Electronic Thesis or Diss., CY Cergy Paris Université, 2023. http://www.theses.fr/2023CYUN1190.
Анотація:
La détection d'observation s'écartant d'un cas de référence est cruciale dans de nombreuses applications. Cette problématique est présente dans la détection de fraudes, l'imagerie médicale, voire même la surveillance vidéo avec des données allant d'image aux sons. La détection d'anomalie profonde a été introduite dans cette optique, en modélisant la classe normale et en considérant tout ce qui est significativement différent comme étant anormal. Dans la mesure où la classe anormale n'est pas bien définie, une classification binaire classique manquerait de robustesse et de fiabilité sur des données hors de son domaine d'apprentissage. Néanmoins, les approches de détection d'anomalies les plus performantes se généralisent encore mal à différents types d'anomalies. Aucune méthode ne permet de simultanément détecter des anomalies d'objets à grande échelle, et des anomalies locales à petite échelle.Dans ce contexte, nous introduisons un premier détecteur d'anomalies plus générique par tâche prétexte. Le modèle, nommé OC-MQ, calcule un score d'anomalie en apprenant à résoudre une tâche prétexte complexe sur la classe normale. La tâche prétexte est composée de plusieurs sous-tâches, séparées en tâche discriminatives et génératives, lui permettant de capturer une grande variété de caractéristiques visuelles.Néanmoins, un ensemble de données d'anomalies supplémentaires est en pratique souvent disponible. Dans cette optique, nous explorons deux approches intégrant des données d'anomalie afin de mieux traiter les cas limites. Tout d'abord, nous généralisons le concept de tâche de prétexte au cas semi-supervisé en apprenant aussi dynamiquement la tâche de prétexte avec des échantillons normaux et anormaux. Nous proposons les modèles SadTPS et SadRest, qui apprennent respectivement une tâche prétexte de reconnaissance de TPS et une tâche de restauration d'image. De plus, nous présentons un nouveau modèle de distance d'anomalie, SadCLR, où l'entraînement est stabilisé par une régularisation contrastive sur la direction des représentations apprises. Nous enrichissons davantage les anomalies existantes en générant plusieurs types de pseudo-anomalies.Enfin, nous prolongeons les deux approches précédentes pour les rendre utilisables avec ou sans données d'anomalies. Premièrement, nous introduisons le modèle AnoMem, qui mémorise un ensemble de prototypes normaux à plusieurs échelles en utilisant des couches de Hopfield modernes. Des estimateurs de distance d'anomalie sont ensuite appris sur les disparités entre l'entrée observée et les prototypes normaux. Deuxièmement, nous reformulons les tâches prétextes apprenables afin qu'elles soient apprises uniquement à partir d'échantillons normaux. Notre modèle proposé, HEAT, apprend de manière adverse la tâche prétexte afin de maintenir de bonnes performance sur les échantillons normaux, tout en échouant sur les anomalies. De plus, nous choisissons la distance de Busemann, récemment proposée dans le modèle du disque de Poincaré, pour calculer le score d'anomalie.Des évaluations approfondies sont réalisées pour chaque méthode proposée, incluant des anomalies grossières, fines ou locales avec comme application l'antifraude visage. Les résultats obtenus dépassant l'état de l'art démontrent le succès de nos méthodesDetecting observations straying apart from a baseline case is becoming increasingly critical in many applications. It is found in fraud detection, medical imaging, video surveillance or even in manufacturing defect detection with data ranging from images to sound. Deep anomaly detection was introduced to tackle this challenge by properly modeling the normal class, and considering anything significantly different as anomalous. Given the anomalous class is not well-defined, classical binary classification will not be suitable and lack robustness and reliability outside its training domain. Nevertheless, the best-performing anomaly detection approaches still lack generalization to different types of anomalies. Indeed, each method is either specialized on high-scale object anomalies or low-scale local anomalies.In this context, we first introduce a more generic one-class pretext-task anomaly detector. The model, named OC-MQ, computes an anomaly score by learning to solve a complex pretext task on the normal class. The pretext task is composed of several sub-tasks allowing it to capture a wide variety of visual cues. More specifically, our model is made of two branches each representing discriminative and generative tasks.Nevertheless, an additional anomalous dataset is in reality often available in many applications and can provide harder edge-case anomalous examples. In this light, we explore two approaches for outlier-exposure. First, we generalize the concept of pretext task to outlier-exposure by dynamically learning the pretext task itself with normal and anomalous samples. We propose two the models SadTPS and SadRest that respectively learn a discriminative pretext task of thin plate transform recognition and generative task of image restoration. In addition, we present a new anomaly-distance model SadCLR, where the training of previously unreliable anomaly-distance models is stabilized by adding contrastive regularization on the representation direction. We further enrich existing anomalies by generating several types of pseudo-anomalies.Finally, we extend the two previous approaches to be usable in both one-class and outlier-exposure setting. Firstly, we introduce the AnoMem model which memorizes a set of multi-scale normal prototypes by using modern Hopfield layers. Anomaly distance estimators are then fitted on the deviations between the input and normal prototypes in a one-class or outlier-exposure manner. Secondly, we generalize learnable pretext tasks to be learned only using normal samples. Our proposed model HEAT adversarially learns the pretext task to be just challenging enough to keep good performance on normal samples, while failing on anomalies. Besides, we choose the recently proposed Busemann distance in the hyperbolic Poincaré ball model to compute the anomaly score.Extensive testing was conducted for each proposed method, varying from coarse and subtle style anomalies to a fraud detection dataset of face presentation attacks with local anomalies. These tests yielded state-of-the-art results, showing the significant success of our methods
32
Guerreiro, Lucas [UNESP]. "Aprendizado semi-supervisionado utilizando modelos de caminhada de partículas em grafos." Universidade Estadual Paulista (UNESP), 2017. http://hdl.handle.net/11449/151923.
Анотація:
Submitted by Lucas Guerreiro null (lucasg@rc.unesp.br) on 2017-10-16T22:03:24Z
No. of bitstreams: 1
LucasGuerreiro_dissertacao.pdf: 2072249 bytes, checksum: 03cb08b42175616dd567a364cf201bcd (MD5)Approved for entry into archive by Monique Sasaki (sayumi_sasaki@hotmail.com) on 2017-10-18T18:42:00Z (GMT) No. of bitstreams: 1 guerreiro_l_me_sjrp.pdf: 2072249 bytes, checksum: 03cb08b42175616dd567a364cf201bcd (MD5)
Made available in DSpace on 2017-10-18T18:42:00Z (GMT). No. of bitstreams: 1 guerreiro_l_me_sjrp.pdf: 2072249 bytes, checksum: 03cb08b42175616dd567a364cf201bcd (MD5) Previous issue date: 2017-09-01
O Aprendizado de Máquina é uma área que vem crescendo nos últimos anos e é um dos destaques dentro do campo de Inteligência Artificial. Atualmente, uma das subáreas mais estudadas é o Aprendizado Semi-Supervisionado, principalmente pela sua característica de ter um menor custo na rotulação de dados de exemplo. A categoria de modelos baseados em grafos é a mais ativa nesta subárea, fazendo uso de estruturas de redes complexas. O algoritmo de competição e cooperação entre partículas é uma das técnicas deste domínio. O algoritmo provê acurácia de classificação compatível com a de algoritmos do estado da arte, e oferece um custo computacional inferior à maioria dos métodos da mesma categoria. Neste trabalho é apresentado um estudo sobre Aprendizado Semi-Supervisionado, com ênfase em modelos baseados em grafos e, em particular, no Algoritmo de Competição e Cooperação entre Partículas (PCC). O objetivo deste trabalho é propor um novo algoritmo de competição e cooperação entre partículas baseado neste modelo, com mudanças na caminhada pelo grafo, com informações de dominância sendo mantidas nas arestas ao invés dos nós; as quais possam melhorar a acurácia de classificação ou ainda o tempo de execução em alguns cenários. É proposta também uma metodologia de avaliação da rede obtida com o modelo de competição e cooperação entre partículas, para se identificar a melhor métrica de distância a ser aplicada em cada caso. Nos experimentos apresentados neste trabalho, pode ser visto que o algoritmo proposto teve melhor acurácia do que o PCC em algumas bases de dados, enquanto o método de avaliação de métricas de distância atingiu também bom nível de precisão na maioria dos casos.
Machine Learning is an increasing area over the last few years and it is one of the highlights in Artificial Intelligence area. Nowadays, one of the most studied areas is Semi-supervised learning, mainly due to its characteristic of lower cost in labeling sample data. The most active category in this subarea is that of graph-based models, using complex networks concepts. The Particle Competition and Cooperation in Networks algorithm (PCC) is one of the techniques in this field. The algorithm provides accuracy compatible with state of the art algorithms, and it presents a lower computational cost when compared to most techniques in the same category. In this project, it is presented a research about semi-supervised learning, with focus on graphbased models and, in special, the Particle Competition and Cooperation in Networks algorithm. The objective of this study is to base proposals of new particle competition and cooperation algorithms based on this model, with new dynamics on the graph walking, keeping dominance information on the edges instead of the nodes; which may improve the accuracy classification or yet the runtime in some situations. It is also proposed a method of evaluation of the network built with the Particle Competition and Cooperation model, in order to infer the best distance metric to be used in each case. In the experiments presented in this work, it can be seen that the proposed algorithm presented better accuracy when compared to the PCC for some datasets, while the proposed distance metrics evaluation achieved a high precision level in most cases.
33
Oliveira, Clayton Silva. "Classificadores baseados em vetores de suporte gerados a partir de dados rotulados e não-rotulados." Universidade de São Paulo, 2006. http://www.teses.usp.br/teses/disponiveis/3/3152/tde-22072007-192518/.
Анотація:
Treinamento semi-supervisionado é uma metodologia de aprendizado de máquina que conjuga características de treinamento supervisionado e não-supervisionado. Ela se baseia no uso de bases semi-rotuladas (bases contendo dados rotulados e não-rotulados) para o treinamento de classificadores. A adição de dados não-rotulados, mais baratos e geralmente disponíveis em maior quantidade do que os dados rotulados, pode aumentar o desempenho e/ou baratear o custo de treinamento desses classificadores (a partir da diminuição da quantidade de dados rotulados necessários). Esta dissertação analisa duas estratégias para se executar treinamento semi-supervisionado, especificamente em Support Vector Machines (SVMs): formas direta e indireta. A estratégia direta é atualmente mais conhecida e estudada, e permite o uso de dados rotulados e não-rotulados, ao mesmo tempo, em tarefas de aprendizagem de classificadores. Entretanto, a inclusão de muitos dados não-rotulados pode tornar o treinamento demasiadamente lento. Já a estratégia indireta é mais recente, sendo capaz de agregar os benefícios do treinamento semi-supervisionado direto com tempos menores para o aprendizado de classificadores. Esta opção utiliza os dados não-rotulados para pré-processar a base de dados previamente à tarefa de aprendizagem do classificador, permitindo, por exemplo, a filtragem de eventuais ruídos e a reescrita da base em espaços de variáveis mais convenientes. Dentro do escopo da forma indireta, está a principal contribuição dessa dissertação: idealização, implementação e análise do algoritmo split learning. Foram obtidos ótimos resultados com esse algoritmo, que se mostrou eficiente em treinar SVMs de melhor desempenho e em períodos menores a partir de bases semi-rotuladas.Semi-supervised learning is a machine learning methodology that mixes features of supervised and unsupervised learning. It allows the use of partially labeled databases (databases with labeled and unlabeled data) to train classifiers. The addition of unlabeled data, which are cheaper and generally more available than labeled data, can enhance the performance and/or decrease the costs of learning such classifiers (by diminishing the quantity of required labeled data). This work analyzes two strategies to perform semi-supervised learning, specifically with Support Vector Machines (SVMs): direct and indirect concepts. The direct strategy is currently more popular and studied; it allows the use of labeled and unlabeled data, concomitantly, in learning classifiers tasks. However, the addition of many unlabeled data can lead to very long training times. The indirect strategy is more recent; it is able to attain the advantages of the direct semi-supervised learning with shorter training times. This alternative uses the unlabeled data to pre-process the database prior to the learning task; it allows denoising and rewriting the data in better feature espaces. The main contribution of this Master thesis lies within the indirect strategy: conceptualization, experimentation, and analysis of the split learning algorithm, that can be used to perform indirect semi-supervised learning using SVMs. We have obtained promising empirical results with this algorithm, which is efficient to train better performance SVMs in shorter times from partially labeled databases.
34
Bharti, Pratool. "Context-based Human Activity Recognition Using Multimodal Wearable Sensors." Scholar Commons, 2017. http://scholarcommons.usf.edu/etd/7000.
Анотація:
In the past decade, Human Activity Recognition (HAR) has been an important part of the regular day to day life of many people. Activity recognition has wide applications in the field of health care, remote monitoring of elders, sports, biometric authentication, e-commerce and more. Each HAR application needs a unique approach to provide solutions driven by the context of the problem. In this dissertation, we are primarily discussing two application of HAR in different contexts. First, we design a novel approach for in-home, fine-grained activity recognition using multimodal wearable sensors on multiple body positions, along with very small Bluetooth beacons deployed in the environment. State-of-the-art in-home activity recognition schemes with wearable devices are mostly capable of detecting coarse-grained activities (sitting, standing, walking, or lying down), but cannot distinguish complex activities (sitting on the floor versus on the sofa or bed). Such schemes are not effective for emerging critical healthcare applications – for example, in remote monitoring of patients with Alzheimer's disease, Bulimia, or Anorexia – because they require a more comprehensive, contextual, and fine-grained recognition of complex daily user activities. Second, we introduced Watch-Dog – a self-harm activity recognition engine, which attempts to infer self-harming activities from sensing accelerometer data using wearable sensors worn on a subject's wrist. In the United States, there are more than 35,000 reported suicides with approximately 1,800 of them being psychiatric inpatients every year. Staff perform intermittent or continuous observations in order to prevent such tragedies, but a study of 98 articles over time showed that 20% to 62% of suicides happened while inpatients were on an observation schedule. Reducing the instances of suicides of inpatients is a problem of critical importance to both patients and healthcare providers. Watch-dog uses supervised learning algorithm to model the system which can discriminate the harmful activities from non-harmful activities. The system is not only very accurate but also energy efficient. Apart from these two HAR systems, we also demonstrated the difference in activity pattern between elder and younger age group. For this experiment, we used 5 activities of daily living (ADL). Based on our findings we recommend that a context aware age-specific HAR model would be a better solution than all age-mixed models. Additionally, we find that personalized models for each individual elder person perform better classification than mixed models.
35
Jones, Joshua K. "Empirically-based self-diagnosis and repair of domain knowledge." Diss., Georgia Institute of Technology, 2009. http://hdl.handle.net/1853/33931.
Анотація:
In this work, I view incremental experiential learning in intelligent software agents as progressive agent self-adaptation. When an agent produces an incorrect behavior, then it may reflect on, and thus diagnose and repair, the reasoning and knowledge that produced the incorrect behavior. In particular, I focus on the self-diagnosis and self-repair of an agent's domain knowledge. The implementation of systems with the capability to self-diagnose and self-repair involves building both reasoning processes capable of such learning and knowledge representations capable of supporting those reasoning processes. The core issue my dissertation addresses is: what kind of metaknowledge (knowledge about knowledge) may enable the agent to diagnose faults in its domain knowledge? In providing a solution to this issue, the central contribution of this research is a theory of the kind of metaknowledge that enables a system to reason about and adapt its conceptual knowledge. For this purpose, I propose a representation that explicitly encodes metaknowledge in the form of procedures called Empirical Verification Procedures (EVPs). In the proposed knowledge representation, an EVP is associated with each concept within the agent's domain knowledge. Each EVP explicitly semantically grounds the associated concept in the agent's perception, and can thus be used as a test to determine the validity of knowledge of that concept during diagnosis.
I present the formal and empirical evaluation of a system, Augur, that makes use of EVP metaknowledge to adapt its own domain knowledge in the context of a particular subclass of classification problem that I call compositional classification, in which the overall classification task can be broken into a hierarchically organized set of subtasks. I hypothesize that EVP metaknowledge will enable a system to automatically adapt its knowledge in two ways: first, by adjusting the ways that inputs are categorized by a concept, in accordance with semantics fixed by an associated EVP; and second, by adjusting the semantics of concepts themselves when they fail to contribute appropriately to system goals. The latter adaptation is realized by altering the EVP associated with the concept in question. I further hypothesize that the semantic grounding of domain concepts in perception through the use of EVPs will increase the generalization power of a learner that operates over those concepts, and thus make learning more efficient. Beyond the support of these hypotheses, I also present results pertinent to the understanding of learning in compositional classification settings using structured knowledge representations.
36
IAPAOLO, FABIO. "De-Individuation of the Modern Subject in the Age of Artificial Intelligence. The Case of Self-Driving Cars and Algorithms for Decision Making." Doctoral thesis, Politecnico di Torino, 2021. http://hdl.handle.net/11583/2875755.
37
Chandra, Nagasai. "Node Classification on Relational Graphs using Deep-RGCNs." DigitalCommons@CalPoly, 2021. https://digitalcommons.calpoly.edu/theses/2265.
Анотація:
Knowledge Graphs are fascinating concepts in machine learning as they can hold usefully structured information in the form of entities and their relations. Despite the valuable applications of such graphs, most knowledge bases remain incomplete. This missing information harms downstream applications such as information retrieval and opens a window for research in statistical relational learning tasks such as node classification and link prediction. This work proposes a deep learning framework based on existing relational convolutional (R-GCN) layers to learn on highly multi-relational data characteristic of realistic knowledge graphs for node property classification tasks. We propose a deep and improved variant, Deep-RGCNs, with dense and residual skip connections between layers. These skip connections are known to be very successful with popular deep CNN-architectures such as ResNet and DenseNet. In our experiments, we investigate and compare the performance of Deep-RGCN with different baselines on multi-relational graph benchmark datasets, AIFB and MUTAG, and show how the deep architecture boosts the performance in the task of node property classification. We also study the training performance of Deep-RGCNs (with N layers) and discuss the gradient vanishing and over-smoothing problems common to deeper GCN architectures.
38
Matsubara, Edson Takashi. "Relações entre ranking, análise ROC e calibração em aprendizado de máquina." Universidade de São Paulo, 2008. http://www.teses.usp.br/teses/disponiveis/55/55134/tde-04032009-114050/.
Анотація:
Aprendizado supervisionado tem sido principalmente utilizado para classificação. Neste trabalho são mostrados os benefícios do uso de rankings ao invés de classificação de exemplos isolados. Um rankeador é um algoritmo que ordena um conjunto de exemplos de tal modo que eles são apresentados do exemplo de maior para o exemplo de menor expectativa de ser positivo. Um ranking é o resultado dessa ordenação. Normalmente, um ranking é obtido pela ordenação do valor de confiança de classificação dado por um classificador. Este trabalho tem como objetivo procurar por novas abordagens para promover o uso de rankings. Desse modo, inicialmente são apresentados as diferenças e semelhanças entre ranking e classificação, bem como um novo algoritmo de ranking que os obtém diretamente sem a necessidade de obter os valores de confiança de classificação, esse algoritmo é denominado de LEXRANK. Uma área de pesquisa bastante importante em rankings é a análise ROC. O estudo de árvores de decisão e análise ROC é bastante sugestivo para o desenvolvimento de uma visualização da construção da árvore em gráficos ROC. Para mostrar passo a passo essa visualização foi desenvolvido uma sistema denominado PROGROC. Ainda do estudo de análise ROC, foi observado que a inclinação (coeficiente angular) dos segmentos que compõem o fecho convexo de curvas ROC é equivalente a razão de verossimilhança que pode ser convertida para probabilidades. Essa conversão é denominada de calibração por fecho convexo de curvas ROC que coincidentemente é equivalente ao algoritmo PAV que implementa regressão isotônica. Esse método de calibração otimiza Brier Score. Ao explorar essa medida foi encontrada uma relação bastante interessante entre Brier Score e curvas ROC. Finalmente, também foram explorados os rankings construídos durante o método de seleção de exemplos do algoritmo de aprendizado semi-supervisionado multi-descrição CO-TRAININGSupervised learning has been used mostly for classification. In this work we show the benefits of a welcome shift in attention from classification to ranking. A ranker is an algorithm that sorts a set of instances from highest to lowest expectation that the instance is positive, and a ranking is the outcome of this sorting. Usually a ranking is obtained by sorting scores given by classifiers. In this work, we are concerned about novel approaches to promote the use of ranking. Therefore, we present the differences and relations between ranking and classification followed by a proposal of a novel ranking algorithm called LEXRANK, whose rankings are derived not from scores, but from a simple ranking of attribute values obtained from the training data. One very important field which uses rankings as its main input is ROC analysis. The study of decision trees and ROC analysis suggested an interesting way to visualize the tree construction in ROC graphs, which has been implemented in a system called PROGROC. Focusing on ROC analysis, we observed that the slope of segments obtained from the ROC convex hull is equivalent to the likelihood ratio, which can be converted into probabilities. Interestingly, this ROC convex hull calibration method is equivalent to Pool Adjacent Violators (PAV). Furthermore, the ROC convex hull calibration method optimizes Brier Score, and the exploration of this measure leads us to find an interesting connection between the Brier Score and ROC Curves. Finally, we also investigate rankings build in the selection method which increments the labelled set of CO-TRAINING, a semi-supervised multi-view learning algorithm
39
Oliverio, Vinicius. "Detecção de contradições em um sistema de aprendizado sem fim." Universidade Federal de São Carlos, 2012. https://repositorio.ufscar.br/handle/ufscar/505.
Анотація:
Made available in DSpace on 2016-06-02T19:05:58Z (GMT). No. of bitstreams: 1
4650.pdf: 1170921 bytes, checksum: a9b0c215ae5f2804a8bbdd8bd2267ca8 (MD5)
Previous issue date: 2012-06-29Universidade Federal de Sao Carlos
NELL (Never Ending Language Learning) is a system that seeks to learn in an infinite way, extracting structured information from unstructured web pages using the semi-supervised learning paradigm as one of its basic principles. The Read the Web (RTW) project is the project where the NELL system is contained, actually it consists of 5 modules, all of them working independently where one of the modules is called Rule Learner (RL). The RL is responsible for inducing first order rules, which are used by the system to identify patterns in the knowledge generated by the other four components of the system. These rules are induced and then represented in a syntax that has Horn clauses as base. These rules can present contradictions, and in this context this paper proposes investigate, develop and implement methods to detect and solve these contradictions so that the system can learn in a more efficient way
O NELL (Never Ending Language Learning) é um sistema que busca aprender de uma maneira contínua, extraindo informação estruturada de páginas web desestruturadas utilizando o paradigma de aprendizagem semissupervisionado como um de seus princípios básicos. O Read the Web (RTW) é o projeto no qual o sistema NELL se insere. Atualmente o NELL possui cinco módulos, todos eles trabalhando independentemente onde um desses módulos é chamado Rule Learner (RL). O RL é responsável por induzir regras de primeira ordem, as quais são utilizadas pelo sistema para identificar padrões presentes no conhecimento gerado pelos outros quatro componentes do sistema. Estas regras são induzidas e, na sequência, representadas através de uma sintaxe que tem cláusulas de Horn como base. Tais regras podem apresentar contradições, e neste contexto o presente trabalho propõe a investigação, desenvolvimento e implementação de métodos para detectar e resolver estas contradições de maneira a fazer a aprendizagem mais eficiente.
40
Gal, Jocelyn. "Application d’algorithmes de machine learning pour l’exploitation de données omiques en oncologie." Electronic Thesis or Diss., Université Côte d'Azur (ComUE), 2019. http://theses.univ-cotedazur.fr/2019AZUR6026.
Анотація:
Le développement de l’informatique en médecine et en biologie a permis de générer un grand volume de données. La complexité et la quantité d’informations à intégrer lors d’une prise de décision médicale ont largement dépassé les capacités humaines. Ces informations comprennent des variables démographiques, cliniques ou radiologiques mais également des variables biologiques et en particulier omiques (génomique, protéomique, transcriptomique et métabolomique) caractérisées par un grand nombre de variables mesurées relativement au faible nombre de patients. Leur analyse représente un véritable défi dans la mesure où elles sont fréquemment « bruitées » et associées à des situations de multi-colinéarité. De nos jours, la puissance de calcul permet d'identifier des modèles cliniquement pertinents parmi cet ensemble de données en utilisant des algorithmes d’apprentissage automatique. A travers cette thèse, notre objectif est d’appliquer des méthodes d’apprentissage supervisé et non supervisé, à des données biologiques de grande dimension, dans le but de participer à l’optimisation de la classification et de la prise en charge thérapeutique des patients atteints de cancers. La première partie de ce travail consiste à appliquer une méthode d’apprentissage supervisé à des données d’immunogénétique germinale pour prédire l’efficacité thérapeutique et la toxicité d’un traitement par inhibiteur de point de contrôle immunitaire. La deuxième partie compare différentes méthodes d’apprentissage non supervisé permettant d’évaluer l’apport de la métabolomique dans le diagnostic et la prise en charge des cancers du sein en situation adjuvante. Enfin la troisième partie de ce travail a pour but d’exposer l’apport que peuvent présenter les essais thérapeutiques simulés en recherche biomédicale. L’application des méthodes d’apprentissage automatique en oncologie offre de nouvelles perspectives aux cliniciens leur permettant ainsi de poser des diagnostics plus rapidement et plus précisément, ou encore d’optimiser la prise en charge thérapeutique en termes d’efficacité et de toxicitéThe development of computer science in medicine and biology has generated a large volume of data. The complexity and the amount of information to be integrated for optimal decision-making in medicine have largely exceeded human capacities. These data includes demographic, clinical and radiological variables, but also biological variables and particularly omics (genomics, proteomics, transcriptomics and metabolomics) characterized by a large number of measured variables relatively to a generally small number of patients. Their analysis represents a real challenge as they are frequently "noisy" and associated with situations of multi-colinearity. Nowadays, computational power makes it possible to identify clinically relevant models within these sets of data by using machine learning algorithms. Through this thesis, our goal is to apply supervised and unsupervised learning methods, to large biological data, in order to participate in the optimization of the classification and therapeutic management of patients with various types of cancer. In the first part of this work a supervised learning method is applied to germline immunogenetic data to predict the efficacy and toxicity of immune checkpoint inhibitor therapy. In the second part, different unsupervised learning methods are compared to evaluate the contribution of metabolomics in the diagnosis and management of breast cancer. Finally, the third part of this work aims to expose the contribution that simulated therapeutic trials can make in biomedical research. The application of machine learning methods in oncology offers new perspectives to clinicians allowing them to make diagnostics faster and more accurately, or to optimize therapeutic management in terms of efficacy and toxicity
41
Sörman, Paulsson Elsa. "Evaluation of In-Silico Labeling for Live Cell Imaging." Thesis, Umeå universitet, Institutionen för fysik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-180590.
Анотація:
Today new drugs are tested on cell cultures in wells to minimize time, cost, andanimal testing. The cells are studied using microscopy in different ways and fluorescentprobes are used to study finer details than the light microscopy can observe.This is an invasive method, so instead of molecular analysis, imaging can be used.In this project, phase-contrast microscopy images of cells together with fluorescentmicroscopy images were used. We use Machine Learning to predict the fluorescentimages from the light microscopy images using a strategy called In-Silico Labeling.A Convolutional Neural Network called U-Net was trained and showed good resultson two different datasets. Pixel-wise regression, pixel-wise classification, andimage classification with one cell in each image was tested. The image classificationwas the most difficult part due to difficulties assigning good quality labels tosingle cells. Pixel-wise regression showed the best result.
42
Krundel, Ludovic. "On microelectronic self-learning cognitive chip systems." Thesis, Loughborough University, 2016. https://dspace.lboro.ac.uk/2134/21804.
Анотація:
After a brief review of machine learning techniques and applications, this Ph.D. thesis examines several approaches for implementing machine learning architectures and algorithms into hardware within our laboratory. From this interdisciplinary background support, we have motivations for novel approaches that we intend to follow as an objective of innovative hardware implementations of dynamically self-reconfigurable logic for enhanced self-adaptive, self-(re)organizing and eventually self-assembling machine learning systems, while developing this new particular area of research. And after reviewing some relevant background of robotic control methods followed by most recent advanced cognitive controllers, this Ph.D. thesis suggests that amongst many well-known ways of designing operational technologies, the design methodologies of those leading-edge high-tech devices such as cognitive chips that may well lead to intelligent machines exhibiting conscious phenomena should crucially be restricted to extremely well defined constraints. Roboticists also need those as specifications to help decide upfront on otherwise infinitely free hardware/software design details. In addition and most importantly, we propose these specifications as methodological guidelines tightly related to ethics and the nowadays well-identified workings of the human body and of its psyche.
43
Andersson, Melanie, Arvola Maja, and Sara Hedar. "Sketch Classification with Neural Networks : A Comparative Study of CNN and RNN on the Quick, Draw! data set." Thesis, Uppsala universitet, Institutionen för teknikvetenskaper, 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-353504.
Анотація:
The aim of the study is to apply and compare the performance of two different types of neural networks on the Quick, Draw! dataset and from this determine whether interpreting the sketches as sequences gives a higher accuracy than interpreting them as pixels. The two types of networks constructed were a recurrent neural network (RNN) and a convolutional neural network (CNN). The networks were optimised and the final architectures included five layers. The final evaluation accuracy achieved was 94.2% and 92.3% respectively, leading to the conclusion that the sequential interpretation of the Quick, Draw! dataset is favourable.
44
Di, Marco Lionel. "Récit d'ingénierie pédagogique en santé à l'usage de l'enseignant connecté Does the acceptance of hybrid learning affect learning approaches in France? Blended Learning for French Health Students: Does Acceptance of a Learning Management System Influence Students’ Self-Efficacy?" Thesis, Université Grenoble Alpes, 2020. http://www.theses.fr/2020GRALS024.
Анотація:
Contexte. L’objectif général de ce travail de thèse était d’évaluer une méthode pédagogique hybride utilisant un environnement informatique pour l’apprentissage humain (EIAH) dans le cadre de la formation des professionnels de santé. Trois questions de recherche se sont ainsi succédé. L’acceptabilité de l’apprentissage hybride influe-t-elle sur les stratégies et approches d’apprentissage des étudiants ? L’acceptabilité d’un EIAH influe-t-elle sur le sentiment d’auto-efficacité des étudiants ? Quelles caractéristiques d’un cours dématérialisé fait varier l’attention des étudiants ?Matériels & Méthodes. Deux études observationnelles quantitatives ont été réalisées, ainsi qu’une expérimentation observationnelle en simple insu couplée à une analyse qualitative, auprès de différentes promotions d’étudiants sages-femmes de l’UFR de Médecine de l’université Grenoble-Alpes.Résultats. Les étudiants ont des approches et stratégies d’apprentissage adaptées malgré l’utilisation d’une méthode pédagogique hybride qu’ils rejettent ; il n’existe pas de corrélation entre une mauvaise acceptabilité de l’EIAH et différentes sphères du sentiment d’auto-efficacité des étudiants ; enfin la variabilité d’attention déclarée par les étudiants varie selon certains facteurs communs à ceux détectés par une intelligence artificielle (type de langage, durée des diapositives…).Discussion. Les validités interne et externe de ces travaux permettent de mettre en avant les liens étroits entre intérêt, motivation, engagement par identification et attention. Il est ainsi possible de proposer des principes d’ingénierie pédagogique en santé dans le cadre de la création des cours dématérialisés axés sur le contenu, la forme et le type de capsule de connaissances. Finalement, l’enseignant en santé doit surtout être « connecté » aux étudiants, pour que les évolutions techniques s’adaptent à ses besoinsBackground. The general objective of this thesis was to evaluate a hybrid pedagogical method using an integrated learning environment (ILE) in the training of health professionals. Three research questions followed one after the other. Does the acceptability of blended learning affect students' learning strategies and learning approaches? Does the acceptability of an ILE affect students' self-efficacy? What characteristics of a dematerialised course make students' attention variable?Materials & Methods. We carried out 2 quantitative observational studies, as well as a single-blind observational experiment coupled with a qualitative analysis, with different classes of midwifery students of Grenoble-Alpes University Faculty of Medicine.Results. Students have suited learning approaches and strategies despite the use of a hybrid teaching method which they reject; there is no correlation between poor acceptability of the ILE and different spheres of students' self-efficacy; and the variability of attention declared by students varies according to certain factors common to those detected by artificial intelligence (type of language, slide duration…).Discussion. The internal and external validities of this work highlight the close links between interest, motivation, engagement by identification, and attention. It is thus possible to put forward principles of pedagogical engineering in health within the framework of dematerialized courses focusing on the content, form and type of knowledge capsule. Finally, the health teacher must above all be “connected to” the students, so that technical developments can be adapted to their needs
45
Albilani, Mohamad. "Neuro-symbolic deep reinforcement learning for safe urban driving using low-cost sensors." Electronic Thesis or Diss., Institut polytechnique de Paris, 2024. http://www.theses.fr/2024IPPAS008.
Анотація:
La recherche effectuée dans cette thèse concerne le domaine de la conduite urbaine sûre, en utilisant des méthodes de fusion de capteurs et d'apprentissage par renforcement pour la perception et le contrôle des véhicules autonomes (VA). L'évolution généralisée des technologies d'apprentissage automatique ont principalement propulsé la prolifération des véhicules autonomes ces dernières années. Cependant, des progrès substantiels sont nécessaires avant d'atteindre une adoption généralisée par le grand public. Pour accomplir son automatisation, les véhicules autonomes nécessitent l'intégration d'une série de capteurs coûteux (e.g. caméras, radars, LiDAR et capteurs à ultrasons). En plus de leur fardeau financier, ces capteurs présentent une sensibilité aux variations telles que la météo, une limitation non partagée par les conducteurs humains qui peuvent naviguer dans des conditions diverses en se fiant à une vision frontale simple. Par ailleurs, l'avènement des algorithmes neuronaux de prise de décision constitue l'intelligence fondamentale des véhicules autonomes. Les solutions d'apprentissage profond par renforcement, facilitant l'apprentissage de la politique du conducteur de bout en bout, ont trouvé application dans des scénarios de conduite élémentaires, englobant des tâches telles que le maintien dans la voie, le contrôle de la direction et la gestion de l'accélération. Cependant, il s'avère que ces algorithmes sont coûteux en temps d'exécution et nécessitent de large ensembles de données pour un entraînement efficace. De plus, la sécurité doit être prise en compte tout au long des phases de développement et de déploiement des véhicules autonomes.La première contribution de cette thèse améliore la localisation des véhicules en fusionnant les mesures des capteurs GPS et IMU avec une adaptation d'un filtre de Kalman, ES-EKF, et une réduction du bruit des mesures IMU. L'algorithme est déployé et testé en utilisant des données de vérité terrain sur un microcontrôleur. La deuxième contribution propose l'algorithme DPPO-IL (Dynamic Proximal Policy Optimization with Imitation Learning), conçu pour faciliter le stationnement automatisé en accordant une attention toute particulière à la sécurité. Cet algorithme apprend à exécuter des manœuvres de stationnement optimales tout en naviguant entre des d'obstacles statiques et dynamiques grâce à un entraînement complet intégrant des données simulées et réelles. La troisième contribution est un framework de conduite urbaine de bout en bout appelé guided hierarchical reinforcement Learning (GHRL). Il intègre des données de vision et de localisation ainsi que des démonstrations d'experts exprimées avec des règles ASP (Answer Set Programming) pour guider la politique d'exploration de l'apprentissage par renforcement hiérarchique et accélérer la convergence de l'algorithme. Lorsqu'une situation critique se produit, le système s'appuie également sur des règles liées à la sécurité pour faire des choix judicieux dans des conditions imprévisibles ou dangereuses. GHRL est évalué sur le jeu de données NoCrash du simulateur Carla et les résultats montrent qu'en incorporant des règles logiques, GHRL obtient de meilleures performances que les algorithmes de l'état de l'artThe research conducted in this thesis is centered on the domain of safe urban driving, employing sensor fusion and reinforcement learning methodologies for the perception and control of autonomous vehicles (AV). The evolution and widespread integration of machine learning technologies have primarily propelled the proliferation of autonomous vehicles in recent years. However, substantial progress is requisite before achieving widespread adoption by the general populace. To accomplish its automation, autonomous vehicles necessitate the integration of an array of costly sensors, including cameras, radars, LiDARs, and ultrasonic sensors. In addition to their financial burden, these sensors exhibit susceptibility to environmental variables such as weather, a limitation not shared by human drivers who can navigate diverse conditions with a reliance on simple frontal vision. Moreover, the advent of decision-making neural network algorithms constitutes the core intelligence of autonomous vehicles. Deep Reinforcement Learning solutions, facilitating end-to-end driver policy learning, have found application in elementary driving scenarios, encompassing tasks like lane-keeping, steering control, and acceleration management. However, these algorithms demand substantial time and extensive datasets for effective training. In addition, safety must be considered throughout the development and deployment phases of autonomous vehicles.The first contribution of this thesis improves vehicle localization by fusing data from GPS and IMU sensors with an adaptation of a Kalman filter, ES-EKF, and a reduction of noise in IMU measurements.This method excels in urban environments marked by signal obstructions and elevated noise levels, effectively mitigating the adverse impact of noise in IMU sensor measurements, thereby maintaining localization accuracy and robustness. The algorithm is deployed and tested employing ground truth data on an embedded microcontroller. The second contribution introduces the DPPO-IL (Dynamic Proximal Policy Optimization with Imitation Learning) algorithm, designed to facilitate end-to-end automated parking while maintaining a steadfast focus on safety. This algorithm acquires proficiency in executing optimal parking maneuvers while navigating static and dynamic obstacles through exhaustive training incorporating simulated and real-world data.The third contribution is an end-to-end urban driving framework called GHRL. It incorporates vision and localization data and expert demonstrations expressed in the Answer Set Programming (ASP) rules to guide the hierarchical reinforcement learning (HRL) exploration policy and speed up the learning algorithm's convergence. When a critical situation occurs, the system relies on safety rules, which empower it to make prudent choices amidst unpredictable or hazardous conditions. GHRL is evaluated on the Carla NoCrash benchmark, and the results show that by incorporating logical rules, GHRL achieved better performance over state-of-the-art algorithms
46
Dsouza, Rodney Gracian. "Deep Learning Based Motion Forecasting for Autonomous Driving." The Ohio State University, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=osu1619139403696822.
47
Giuliani, Luca. "Extending the Moving Targets Method for Injecting Constraints in Machine Learning." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2021. http://amslaurea.unibo.it/23885/.
Анотація:
Informed Machine Learning is an umbrella term that comprises a set of methodologies in which domain knowledge is injected into a data-driven system in order to improve its level of accuracy, satisfy some external constraint, and in general serve the purposes of explainability and reliability. The said topid has been widely explored in the literature by means of many different techniques. Moving Targets is one such a technique particularly focused on constraint satisfaction: it is based on decomposition and bi-level optimization and proceeds by iteratively refining the target labels through a master step which is in charge of enforcing the constraints, while the training phase is delegated to a learner. In this work, we extend the algorithm in order to deal with semi-supervised learning and soft constraints. In particular, we focus our empirical evaluation on both regression and classification tasks involving monotonicity shape constraints. We demonstrate that our method is robust with respect to its hyperparameters, as well as being able to generalize very well while reducing the number of violations on the enforced constraints. Additionally, the method can even outperform, both in terms of accuracy and constraint satisfaction, other state-of-the-art techniques such as Lattice Models and Semantic-based Regularization with a Lagrangian Dual approach for automatic hyperparameter tuning.
48
Ansarnia, Masoomeh. "Development and Test of Computer Vision and Deep Learning Methods for Dynamic Management of Urban Lighting." Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0272.
Анотація:
Cette thèse de doctorat est réalisée dans le cadre d'un contrat de recherche entre la Société Eclatec, concepteur et fabricant français de luminaires urbains, et l'Institut Jean Lamour de Nancy. L'objectif concerne l'amélioration de l'éclairage nocturne des villes tout en réduisant la consommation électrique et la pollution lumineuse. Pour cela une caméra RGB est intégrée à la source lumineuse du lampadaire. Elle constitue le principal point de collecte d'informations. Ce choix a imposé l'utilisation d'une optique très grand angle dont l'axe est faiblement incliné par rapport à la verticale. La configuration adoptée permet d'observer en grande partie la zone éclairée par le luminaire mais se traduit par des images très déformées. A partir de ce système 4 grandes problématiques ont été étudiées. La première concerne la détection vidéo des usagers à proximité du luminaire dans des conditions de très faible éclairement afin d'assurer une gradation dynamique de l'éclairage. Cette détection exploite les modèles de deep learning de la famille Yolo que nous avons ré-entrainés par transfert learning sur une collection d'images spécifiques. Celles-ci ont été relevées en différents points de l'agglomération nancéenne à une hauteur de 6 à 8 m. Pour une scène éclairée sous 10 lux, une ouverture photographique à f/3.5 et une sensibilité fixée à 3200 ISO, le taux de détection des piétons et véhicules est supérieur à 97 %. Le modèle, implanté sur le GPU embarqué NVidia Jetson nano, permet d'atteindre une cadence d'environ 10 FPS qui reste suffisante pour notre application. La deuxième orientation étudie la reconnaissance de l'environnement autour du luminaire au moyen d'une segmentation sémantique de images. Cette segmentation sera exploitée ultérieurement pour adapter la distribution lumineuse de la matrice de leds à la situation urbaine rencontrée. Pour ce faire, nous avons fait appel au réseau de neurones OCR-HRNet qui améliore la segmentation haute résolution par l'ajout d'une représentation contextuelle tenant compte de l'agrégation des pixels. Cette architecture s'avère bien adaptée aux images de surfaces peu homogènes qui caractérisent le sol sous le luminaire. Les résultats montrent une très bonne identification des constructions et des zones végétalisées. La séparation trottoir/route reste encore délicate notamment lorsque les revêtements des voies de circulation présentent des réflectances et textures similaires. Une solution par marquage virtuel à postériori des images améliore sensiblement la précision de la segmentation notamment dans le cas de scènes ensoleillées qui présentent de nombreuses zones d'ombre. Dans un troisième temps nous avons modélisé le système optique afin de permettre une estimation de la position réelle des points du sol à partir de leur image. Une transformation Cam To World simple est proposée. Celle-ci prend en compte les paramètres extrinsèques de la prise de vue (hauteur, pitch et résolution) et la fonction de distorsion de la lentille dont la projection optique est assimilée à une loi équidistante. La précision requise n'étant pas critique, la calibration rigoureuse du système n'a pas été effectuée. Pour une zone d'observation effective de 20 m × 50 m, l'erreur de localisation est de l'ordre du mètre. Enfin nous proposons une piste d'exploitation du parc de luminaires pour l'analyse de la fluidité du trafic routier. La méthode proposée analyse le mouvement apparent des usagers par estimation du flot optique moyen dans chacune des boites englobantes détectées par Yolo. La détermination du flot optique est actuellement réalisée en mode hors ligne par l'algorithme de deep learning FlowNet2. Dans la gamme de 0 à 15 m/s la vitesse estimée du mobile présente une erreur inférieure à 1 m/sThis doctoral thesis has been conducted within the framework of a research contract between the French urban lighting design and manufacturing company, Eclatec, and the Jean Lamour Institute in Nancy. The overarching goal of this research is to enhance nighttime urban lighting while simultaneously reducing electrical consumption and light pollution. To achieve this, an RGB camera is integrated into the streetlamp's light source, serving as the primary data collection point. This choice necessitated the use of a wide-angle lens with a slight vertical tilt in its axis. Although this configuration allows for the observation of a significant portion of the illuminated area, it results in highly distorted images. From this system, four major research challenges were investigated:1. The first challenge concerns video detection of individuals in close proximity to the luminaire under very low lighting conditions, with the aim of achieving dynamic lighting adjustment. This detection relies on deep learning models from the Yolo family, which were fine-tuned through transfer learning using a specific collection of images. These images were captured at various locations in the Nancy metropolitan area, at heights ranging from 6 to 8 meters. Under conditions of 10 lux illumination, an aperture of f/3.5, and a fixed sensitivity of 3200 ISO, the detection rate for pedestrians and vehicles exceeds 97%. The model, implemented on the embedded NVidia Jetson Nano GPU, achieves a frame rate of approximately 10 FPS, which proves adequate for our application. 2. The second research direction explores the recognition of the environment surrounding the luminaire through semantic segmentation of images. This segmentation will subsequently be employed to adapt the light distribution of the LED matrix to the encountered urban scenario. To accomplish this, we employed the OCR-HRNet neural network, which enhances high-resolution segmentation by incorporating contextual representation that considers pixel aggregation. This architecture is well-suited to images of non-uniform surfaces, characteristic of the ground beneath the luminaire. The results demonstrate excellent identification of structures and vegetated areas. However, the distinction between sidewalk and road remains challenging, particularly when road surfaces exhibit similar reflectance and textures. A post-image virtual marking solution significantly improves segmentation accuracy, especially in sunny scenes with numerous shadowed areas. 3. In a third phase, we modeled the optical system to enable the estimation of the real-world positions of ground points based on their images. A simple Cam To World transformation is proposed, accounting for extrinsic parameters of the viewpoint (height, pitch, and resolution), and the lens distortion function, approximated as an equidistant projection law. Given that stringent precision is not critical, a rigorous system calibration was not conducted. For an effective observation zone of 20 m × 50 m, the localization error is on the order of meters. 4. Finally, we propose an avenue for utilizing the lighting infrastructure to analyze traffic flow fluidity. The proposed method analyzes apparent motion of users by estimating the mean optical flow within each bounding box detected by Yolo. Currently, optical flow determination is performed offline using the deep learning algorithm FlowNet2. In the range of 0 to 15 m/s, the estimated speed of the moving object exhibits an error of less than 1 m/s
49
Rabe, Erik, and Zacharias Sundlöf. "Bidragande faktorer till attityder gentemot implementering av AI-styrda fordon." Thesis, Uppsala universitet, Institutionen för informatik och media, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-417547.
Анотація:
Artificiell intelligens är en form av teknik som blir vanligare inom samhället. I takt med att tekniken utvecklas blir även diskussionen inom området mer utvecklad vilket resulterat i att eventuella problem och möjligheter blivit mer tillgänglig information. Det finns en avsaknad av tankar och förväntningar från privatpersoners synvinkel inom ämnet vilket kan ses som negativt då de förväntas vara en majoritet av användarbasen för tekniken. Eftersom denna typ av teknik förutspås ta över ett större ansvar av mänskliga uppgifter är det viktigt att klarlägga olika typer av tillvägagångssätt samt utvecklingsperspektiv i syfte att skapa ett hälsosamt och välfungerande AI-system till respektive områden. Studien syftar till att belysa bidragande faktorer till attityder och åsikter relaterade specifikt till AI-styrda fordon ur privatpersoners perspektiv samt hur dessa kan påverka en eventuell implementering och använder sig av en kvalitativ metod. Den data som används inom arbetet har samlats in via semistrukturerade intervjuer med personer som anmält att de vill delta i studien. Analysen genomförs baserat på innovationsspridningsteorin (IDT) och relevant tidigare forskning för att undersöka vad som påverkar användare att adoptera tekniken eller inte. Faktorer som identifierades vara påverkande för adoptionsprocessen var oro över att tekniken inte skulle fungera på ett kompatibelt sätt med mänskliga värderingar, ett krav på utförlig testning samt möjligheten till att reducera olyckor eller klimatpåverkan relaterat till trafik. Utifrån dessa faktorer härleddes förslag till implementeringsprocesser vilket bestod av expanderande statligt kontrollerad testning inom kollektivtrafiken, tydligt klarlagda strukturella regler och avgränsningar samt ett främjande av de positiva faktorer som möjliggörs av AI-styrda fordon. Detta främjande kan genomföras med en effektiv kommunikation som drar nytta av vår bristfällliga rationella beslutsprocess och använder starka känslomässiga intryck.Artificial intelligence is a form of technology that is becoming increasingly more common within society. As the technology evolves, the discussion within the subject is also increasing which has made information about eventual problems and possibilities more public. There is a shortage of thoughts and expectations from the private individual’s point of view regarding this topic which can be a negative thing due to this group being expected to make up the majority of the technology’s user base. Because this type of technology is predicted to take on a larger responsibility of human tasks it is important to clarify different approaches and development perspectives in order to create a healthy and well-functioning AI-system within respective areas. The study intends to highlight contributing factors to attitudes and opinions specifically related to AI-controlled vehicles from the public's view as well as how these can affect an eventual implementation and is carried out with a qualitative method. The data that is used is gathered through semi-structured interviews with people that expressed interest in participating in the study. The analysis is based on the diffusion of innovations theory (IDT) and relevant earlier research in order to examine what influences users to adopt the technology or not. The factors that were identified to be affecting this process were worry that the technology would not work in a compatible way with human values, a demand for extensive testing as well as the possibility to reduce accidents or the affect on climate related to traffic. Several suggestions for implementation were derived from these factors which consisted of continuous expanded testing within public transport regulated by the state, clear structural rules and limitations as well as a promotion of the positive factors made possible by AI-controlled vehicles. This promotion can be done through effective communication which takes advantage of our flawed rational decision making and uses strong emotional impressions.
50
Zaiem, Mohamed Salah. "Informed Speech Self-supervised Representation Learning." Electronic Thesis or Diss., Institut polytechnique de Paris, 2024. http://www.theses.fr/2024IPPAT009.
Анотація:
L'apprentissage des caractéristiques a été un des principaux moteurs des progrès de l'apprentissage automatique. L'apprentissage auto-supervisé est apparu dans ce contexte, permettant le traitement de données non étiquetées en vue d'une meilleure performance sur des tâches faiblement étiquetées. La première partie de mon travail de doctorat vise à motiver les choix dans les pipelines d'apprentissage auto-supervisé de la parole qui apprennent les représentations non supervisées. Dans cette thèse, je montre d'abord comment une fonction basée sur l'indépendance conditionnelle peut être utilisée pour sélectionner efficacement et de manière optimale des tâches de pré-entraînement adaptées à la meilleure performance sur une tâche cible. La deuxième partie de mon travail de doctorat étudie l'évaluation et l'utilisation de représentations auto-supervisées pré-entraînées. J'y explore d'abord la robustesse des benchmarks actuels d'auto-supervision de la parole aux changements dans les choix de modélisation en aval. Je propose, ensuite, de nouvelles approches d'entraînement en aval favorisant l'efficacité et la généralisationFeature learning has been driving machine learning advancement with the recently proposed methods getting progressively rid of handcrafted parts within the transformations from inputs to desired labels. Self-supervised learning has emerged within this context, allowing the processing of unlabeled data towards better performance on low-labeled tasks. The first part of my doctoral work is aimed towards motivating the choices in the speech selfsupervised pipelines learning the unsupervised representations. In this thesis, I first show how conditional-independence-based scoring can be used to efficiently and optimally select pretraining tasks tailored for the best performance on a target task. The second part of my doctoral work studies the evaluation and usage of pretrained self-supervised representations. I explore, first, the robustness of current speech self-supervision benchmarks to changes in the downstream modeling choices. I propose, second, fine-tuning approaches for better efficicency and generalization