Intelligence artificielle et données
488 documents à télécharger gratuitement
Cours, examens, TD, TP et exercices de intelligence artificielle et données. Thèmes couverts : machine learning, apprentissage automatique, big data, data science, fouille de données.
This document introduces a cascaded multi-task framework utilizing convolutional neural networks (CNNs) for joint face detection and alignment in challenging environments with pose variations and occlusions. The methodology involves three stages of CNNs refining candidate windows and producing bounding box and facial landmark positions progressively. It features a novel online hard sample mining strategy to enhance model performance dynamically during training. Experimental results demonstrate superior accuracy and computational efficiency on benchmarks like FDDB, WIDER FACE, and AFLW, surp...
This document provides a comprehensive overview of Hadoop HDFS operations, including starting HDFS, listing files, inserting data, retrieving data, and shutting down the system. It details commands for formatting the file system, managing directories, and transferring files between local systems and HDFS. These operations are essential for effectively utilizing the Hadoop distributed file system.
This document provides an overview of the Hadoop Distributed File System (HDFS), highlighting its architecture and features. It explains the roles of the namenode and datanode, as well as the concept of storage blocks. Additionally, it discusses the goals of HDFS in terms of fault tolerance and efficient data management.
This document outlines the setup process for the Hadoop environment on a Linux operating system. It includes steps for user creation, SSH setup, Java installation, and downloading Hadoop. The guide provides detailed commands to ensure a successful installation of Hadoop.
Hadoop is an Apache open source framework that enables distributed processing of large datasets across clusters of computers using simple programming models. It comprises multiple modules, including Hadoop Common, YARN, HDFS, and MapReduce, which work together to provide a reliable and scalable framework for data management. This document introduces the core components of Hadoop and their functionalities.
This document contains an exam for the Master 1 program in Computer Science at Université Paris 7, focusing on Artificial Intelligence. The exam consists of exercises related to search algorithms and heuristics, including A* search and alpha-beta pruning techniques. Students are required to analyze heuristics for admissibility and dominance, as well as apply search strategies to find optimal paths in given scenarios.
The document explores the concept and structure of expert systems, highlighting their role as tools that replicate human expert reasoning in specific domains. It delves deeper into expert system generators, explaining their composition, such as knowledge bases, inference engines, and interfaces, and reviews notable examples like ADA, Prolog, and Jess. The applications of these technologies, ranging from medicine to industrial systems, are emphasized. Additionally, the document covers the evolution and interoperability of related programming languages like ADA and Prolog in system design.
This article addresses the problem of video segmentation by proposing a novel method that combines active contours with watershed presegmentation. The method aims to refine segmentation results by introducing a force of attraction to the watershed boundaries, enhancing both precision and regularity. The two-step process helps in accurately detecting moving objects in videos, yielding smoother results than existing methods.
Ce document traite de la détection de points d’intérêts dans les images, un processus essentiel dans la vision par ordinateur. Les méthodes de détection sont classées en trois catégories : basées sur les contours, sur l'intensité, et des approches basées sur des modèles. Le détecteur de Moravec est également présenté comme une méthode pour identifier les changements d'intensité dans une zone considérée.
Ce document traite des différentes catégories de raisonnement en intelligence artificielle, notamment le raisonnement déductif, inductif et abductif. Il aborde également la question de l'incertitude et de l'approximativité dans les raisonnements, ainsi que l'importance du temps dans les systèmes d'IA. Enfin, il explore diverses logiques et méthodes pour traiter ces aspects dans le domaine de l'IA.
Ce document explore les notions du raisonnement en intelligence artificielle, en se concentrant sur les symboles et la représentation des connaissances. Il aborde l'historique et les définitions des ressources nécessaires pour un raisonnement efficace. Différents types de raisonnement, ainsi que leurs techniques et défis, sont également discutés.
This document discusses classification methods in data mining, focusing on Naive Bayes with numeric attributes and examples regarding probability calculations. It also addresses the limitations of Naive Bayes with nominal attributes and introduces decision tree algorithms such as ID3, C4.5, and C5.0 for numeric attributes.
This document covers a course on Data Mining focusing on Decision Trees. It includes definitions, examples, and how to construct a decision tree. Key methods of classification such as ID3, C4.5, and C5.0 are also discussed.
This document provides an overview of supervised classification using decision trees. It includes an introduction to the structure of decision trees and details the ID3 algorithm along with its characteristics. Additionally, the document references course notes and literature related to the topic.
This document covers supervised classification methods focusing on representative examples, particularly the k-Nearest Neighbors (kNN) algorithm. It explores the concept of dissimilarity measures and their importance in classification tasks. Practical applications using Weka and reference materials are provided for further study.
This document contains notes from a data mining course focused on supervised classification methods. It includes a discussion of methods like OneR and Naive Bayes, their applications, and exercises involving datasets. Key concepts such as probability estimation in classification are also addressed.
This document presents notes from a course on Data Mining, focusing on supervised classification methods. It covers the basics of classification, including the ZeroR and Naïve Bayes methods. The aim is to explain the process of constructing classifiers and evaluating their performance.
This lecture introduces text mining and Natural Language Processing (NLP), emphasizing their definitions and applications. It addresses the challenges of analyzing natural text data and describes various types of linguistic knowledge essential for processing text. Applications such as language identification, sentiment analysis, and document categorization are discussed.
This document discusses the classification of texts using the Naive Bayes method, focusing on the detection of spam emails. It covers the problem and the data used for analysis, along with the representation of texts and the supervised classification process. The solution includes normalizing the texts, constructing a vocabulary, and extracting attributes for classification.
This document outlines a lab exercise focused on applying the Naïve Bayes method for text classification. It involves working with a dataset of emails to classify them as spam or non-spam. The exercise includes steps for text representation, probability calculation, and classification of new emails.
This document outlines an exam exercise focused on data mining techniques applied to the 'play tennis' dataset. It covers the application of simple classification methods such as ZeroR, OneR, and Naive Bayes using Weka. Students are required to classify instances and evaluate the performance of different classifiers using specified training and testing samples.
The document consists of exercises that explore decision-making models using graphical representations. It includes scenarios where individuals must make strategic decisions based on uncertain outcomes, using concepts from probability and utility theory. The exercises guide the reader through creating decision trees and Bayesian networks for optimization.
This document covers the fundamentals of data mining, including various data types and attributes. It also discusses supervised classification, decision trees, and Bayesian classifiers. Exercises and examples are provided to reinforce the concepts.
The document provides an overview of Artificial Intelligence (AI), its definition, and goals, including modeling human reasoning processes and reproducing human behavior. It highlights the Turing Test and philosophical critiques like the Chinese Room argument, differentiating approaches to evaluating machine intelligence. It reviews AI's historical milestones—from early programs like ELIZA to modern developments in natural language processing, machine learning, and robotics. Additionally, it explores societal applications of AI, such as recommendation systems, automated vehicles, and roboti...