Deep Learning

Closed
Course
fr
French
21 h
This content is rated 0 out of 5
Source
  • From www.fun-mooc.fr
Conditions
  • Free Access
  • Free certificate
More info
  • 35 Sequences
  • Intermediate Level
  • Starts on November 1, 2020
  • Ends on December 12, 2020

Their employees are learning daily with Edflex

  • Safran
  • Air France
  • TotalEnergies
  • Generali
Learn more

Course details

Syllabus

Semaine 1

  • Introduction and Context
  • The Formal Neuron
  • Neural Networks
  • Supervised Machine Learning.
  • Gradient Error Backpropagation Algorithm
  • Training Issues

Semaine 2

  • Introduction to Convolutionnal Neural Networks
  • Convolution
  • Convolution Properties
  • Pooling
  • Convolution Layer
  • Convolution Hierarchies and Convolutional Neural Networks

Semaine 3

  • Modeling Capacities
  • Representation Learning with ConvNets
  • Manifold Untangling and Visualization
  • Deep Learning Weaknesses
  • Success Deep Learning History
  • Deep Learning Winter History

Semaine 4

  • Deep Learning Renewal
  • Modern Convolutional Neural Network Modules
  • Modern Training and Regularization
  • Modern Convolutional Neural Network Architectures
  • Residual Architectures
  • Other Modern Deep Learning Components

Semaine 5

  • Optimization Issues
  • Advanced Optimization
  • Implementations Issues
  • Deep Learning Resources
  • Deep Learning Resources: Keras

Semaine 6

  • Deep Features
  • Transfer Learning
  • Localization and Segmentation
  • Unsupervised Learning
  • Vision and Langage
  • Deep Learning Theory

Prerequisite

Ce cours requiert des connaissances de bases en calcul différentiel et en statistiques, de niveau premier cycle universitaire scientifique.

Instructors

Nicolas Thome
Nicolas Thome est professeur des universités au Conservatoire national des arts et métiers, et chercheur au laboratoire CEDRIC dans l’équipe MSDMA. Ces travaux portent sur l’apprentissage statistique appliqué à des problèmes de compréhension de données multimédia. Il est actuellement impliqué dans de nombreux projets collaboratifs français et internationaux sur le deep learning.

Editor

The Conservatoire national des arts et métiers is a major higher education establishment dedicated to lifelong learning. Created by the Convention in 1794 at the suggestion of Abbé Henri Grégoire "to improve national industry", CNAM is now a public scientific, cultural and professional institution with the status of a grand établissement.

Every year, the public institution, its 28 regional centres and its 150 teaching centres welcome almost 100,000 students (employees, jobseekers, self-employed workers), who come to Cnam to update their knowledge, improve their skills or acquire a diploma, from baccalaureate level to post-graduate and engineering diplomas.

In order to meet the expectations of its customers, Cnam has set itself the target of offering 70% of its training courses in digital format, enabling a variety of learning methods to be used (face-to-face, hybrid, distance learning) and ensuring that its courses are accessible to everyone, anywhere and at any time.

In addition to this objective, the Cnam has included in its master plan the development of massive, open and online courses (MOOC) for the French-speaking public, which will complement its range of degree courses in different ways. The first courses in the fields of management, health, IT and mathematics are already in production.

 

Platform

France Université Numérique is the broadcaster of the online courses of French higher education institutions and their partners.

It operates several platforms of diffusion, of which the best known, FUN MOOC, is the first French-speaking academic platform worldwide. Thanks to many partner institutions, this platform offers a vast catalog of courses enriched daily with various themes and current events.

This content is rated 4.5 out of 5
(no review)
This content is rated 4.5 out of 5
(no review)
Complete this resource to write a review