Global ETD Search

331	Génération de données synthétiques pour l'adaptation hors-domaine non-supervisée en réponse aux questions : méthodes basées sur des règles contre réseaux de neurones Duran, Juan Felipe 02 1900 (has links) Les modèles de réponse aux questions ont montré des résultats impressionnants sur plusieurs ensembles de données et tâches de réponse aux questions. Cependant, lorsqu'ils sont testés sur des ensembles de données hors domaine, la performance diminue. Afin de contourner l'annotation manuelle des données d'entraînement du nouveau domaine, des paires de questions-réponses peuvent être générées synthétiquement à partir de données non annotées. Dans ce travail, nous nous intéressons à la génération de données synthétiques et nous testons différentes méthodes de traitement du langage naturel pour les deux étapes de création d'ensembles de données : génération de questions et génération de réponses. Nous utilisons les ensembles de données générés pour entraîner les modèles UnifiedQA et Bert-QA et nous les testons sur SCIQ, un ensemble de données hors domaine sur la physique, la chimie et la biologie pour la tâche de question-réponse à choix multiples, ainsi que sur HotpotQA, TriviaQA, NatQ et SearchQA, quatre ensembles de données hors domaine pour la tâche de question-réponse. Cette procédure nous permet d'évaluer et de comparer les méthodes basées sur des règles avec les méthodes de réseaux neuronaux. Nous montrons que les méthodes basées sur des règles produisent des résultats supérieurs pour la tâche de question-réponse à choix multiple, mais que les méthodes de réseaux neuronaux produisent généralement des meilleurs résultats pour la tâche de question-réponse. Par contre, nous observons aussi qu'occasionnellement, les méthodes basées sur des règles peuvent compléter les méthodes de réseaux neuronaux et produire des résultats compétitifs lorsqu'on entraîne Bert-QA avec les bases de données synthétiques provenant des deux méthodes. / Question Answering models have shown impressive results in several question answering datasets and tasks. However, when tested on out-of-domain datasets, the performance decreases. In order to circumvent manually annotating training data from the new domain, question-answer pairs can be generated synthetically from unnanotated data. In this work, we are interested in the generation of synthetic data and we test different Natural Language Processing methods for the two steps of dataset creation: question/answer generation. We use the generated datasets to train QA models UnifiedQA and Bert-QA and we test it on SCIQ, an out-of-domain dataset about physics, chemistry, and biology for MCQA, and on HotpotQA, TriviaQA, NatQ and SearchQA, four out-of-domain datasets for QA. This procedure allows us to evaluate and compare rule-based methods with neural network methods. We show that rule-based methods yield superior results for the multiple-choice question-answering task, but neural network methods generally produce better results for the question-answering task. However, we also observe that occasionally, rule-based methods can complement neural network methods and produce competitive results when training Bert-QA with synthetic databases derived from both methods. Intelligence Artificielle Adaptation de domaine Génération automatique de questions Génération automatique de réponses Méthodes basées sur des règles Apprentissage profond Apprentissage non supervisé Automatic question generation Automatic answer generation Methods based on neural networks Rule-based methods Deep learning Unsupervised learning Domain adaptation NLP (Natural Language Processing) Artificial intelligence
332	Malicious Intent Detection Framework for Social Networks Fausak, Andrew Raymond 05 1900 (has links) Many, if not all people have online social accounts (OSAs) on an online community (OC) such as Facebook (Meta), Twitter (X), Instagram (Meta), Mastodon, Nostr. OCs enable quick and easy interaction with friends, family, and even online communities to share information about. There is also a dark side to Ocs, where users with malicious intent join OC platforms with the purpose of criminal activities such as spreading fake news/information, cyberbullying, propaganda, phishing, stealing, and unjust enrichment. These criminal activities are especially concerning when harming minors. Detection and mitigation are needed to protect and help OCs and stop these criminals from harming others. Many solutions exist; however, they are typically focused on a single category of malicious intent detection rather than an all-encompassing solution. To answer this challenge, we propose the first steps of a framework for analyzing and identifying malicious intent in OCs that we refer to as malicious mntent detection framework (MIDF). MIDF is an extensible proof-of-concept that uses machine learning techniques to enable detection and mitigation. The framework will first be used to detect malicious users using solely relationships and then can be leveraged to create a suite of malicious intent vector detection models, including phishing, propaganda, scams, cyberbullying, racism, spam, and bots for open-source online social networks, such as Mastodon, and Nostr. Artificial Intelligence (AI) Machine Learning (ML) Deep Learning (DL) Natural Language Processing (NLP) Cybersecurity Malicious Intent Detection Anomaly Detection Threat Intelligence Behavior Analysis Text Classification Sentiment Analysis Predictive Modeling Neural Networks Supervised Learning Unsupervised Learning Reinforcement Learning Data Mining Feature Extraction Algorithmic Bias Ethics in AI Cyber Threats Intrusion Detection Systems (IDS) Phishing Detection Social Engineering Online Behavior Analysis Artificial Intelligence Computer Science Statistics
333	Reparametrization in deep learning Dinh, Laurent 02 1900 (has links) No description available. Neural networks Deep neural networks Machine learning Deep learning Unsupervised learning Probabilistic modelling Probabilistic models Generative modelling Generative models Generator networks Variational inference Generalization Reparametrization trick Réseaux de neurones Réseaux neuronaux Réseaux de neurones profonds Réseaux neuronaux profonds Apprentissage automatique Apprentissage profond Apprentissage non-supervisé Modélisation probabiliste Modélisation générative Modèles probabilistes Modèles génératifs Réseaux générateurs Inférence variationnelle Généralisation Astuce de la reparamétrisation
334	Dynamics of Forest Ecosystems Under Global Change: Applications of Artificial Intelligence in Mapping, Classification, and Projection Akane Ota Abbasi (17123185) 10 October 2023 (has links) <p dir="ltr">Global forest ecosystems provide essential ecosystem services that contribute to water and climate regulation, food production, recreation, and raw materials. They also serve as crucial habitats for numerous terrestrial species of amphibians, birds, and mammals worldwide. However, recent decades have witnessed unprecedented changes in forest ecosystems due to climate change, shifts in species distribution patterns, increased planted forest areas, and various disturbances such as forest fires, insect infestations, and urbanization. These changes can have far-reaching impacts on ecological networks, human well-being, and the well-being of global forest ecosystems. To address these challenges, I present four studies to quantify forest dynamics through mapping, classification, and projection, using artificial intelligence tools in combination with a vast amount of training data. (I) I present a spatially continuous map of planted forest distribution across East Asia, produced by integrating multiple sources of planted and natural forest data. I found that China contributed 87% of the total planted forest areas in East Asia, most of which are located in the lowland tropical/subtropical regions and Sichuan Basin. I also estimated the dominant genus in each planted forest location. (II) I used continent-wide forest inventory data to compare the range shifts of forest types and their constituent tree species in North America in the past 50 years. I found that forest types shifted more than three times as fast as the average of their constituent tree species. This marked difference was attributable to a predominant positive covariance between tree species ranges and the change of species relative abundance. (III) Based on individual-level field surveys of trees and breeding birds across North America, I characterized New World wood-warbler (<i>Parulidae</i>) species richness and its potential drivers. I identified forest type as the most powerful predictor of New World wood-warbler species richness, which adds valuable evidence to the ongoing physiognomy versus composition debate among ornithologists. (IV) In the appendix, I utilized continent-wide forest inventory data from North America and South America and the combination of supervised and unsupervised machine learning algorithms to produce the first data-driven map of forest types in the Americas. I revealed the distribution of forest types, which are useful for cost-effective forest and biodiversity management and planning. Taken together, these studies provide insight into the dynamics of forest ecosystems at a large geographic scale and have implications for effective decision-making in conservation, management, and global restoration programs in the midst of ongoing global change.</p> Forest biodiversity Forest ecosystems Modelling and simulation Deep learning Neural networks Semi- and unsupervised learning forest dynamics modeling Global Change Climate Change Machine Learning Biodiversity Forest Ecology forest ecosystem modeling Planted Forests Forest type classification Species Distribution Deep Learning Forest Inventory & Analysis Program Forest Inventory Parulidae Species Richness Habitat physiognomy Habitat Heterogeneity Habitat Composition Markowitz portfolio selection
335	EXPLORING GRAPH NEURAL NETWORKS FOR CLUSTERING AND CLASSIFICATION Fattah Muhammad Tahabi (14160375) 03 February 2023 (has links) <p><strong>Graph Neural Networks</strong> (GNNs) have become excessively popular and prominent deep learning techniques to analyze structural graph data for their ability to solve complex real-world problems. Because graphs provide an efficient approach to contriving abstract hypothetical concepts, modern research overcomes the limitations of classical graph theory, requiring prior knowledge of the graph structure before employing traditional algorithms. GNNs, an impressive framework for representation learning of graphs, have already produced many state-of-the-art techniques to solve node classification, link prediction, and graph classification tasks. GNNs can learn meaningful representations of graphs incorporating topological structure, node attributes, and neighborhood aggregation to solve supervised, semi-supervised, and unsupervised graph-based problems. In this study, the usefulness of GNNs has been analyzed primarily from two aspects - <strong>clustering and classification</strong>. We focus on these two techniques, as they are the most popular strategies in data mining to discern collected data and employ predictive analysis.</p> Biomechanical engineering Neural engineering Health promotion Preventative health care Applications in health Spatial data and applications Evolutionary computation Natural language processing Planning and decision making Data engineering and data science Data mining and knowledge discovery Graph, social and multimedia data Information retrieval and web search Knowledge and information management Context learning Deep learning Neural networks Semi- and unsupervised learning Data structures and algorithms Graph neural network Node classification Graph clustering Temporal graphs dynamic graphs NODE2VEC Graph Attention Mechanism Hunting BiLSTM model EHR data colorectal Cancer Cancers Cancer symptoms symptom Symptom cluster studies Coauthorship networks network analysis Word2vec Hierarchical Clustering method Dunn index semantic analysis text mining Natural Language Processing Tool UMLS identifiers umls Clinical Data Management

Page generated in 0.0747 seconds