Neuronal Networks and Deep Learning

CERTIFICATION

This training corresponds to a block of competences for the certificate “Artificial Intelligence Project Manager”, currently under instruction at France Compétences. It will lead to a certification recognized and registered in the Répertoire national des certifications professionnelles (RNCP) and will therefore be eligible for the CPF.

This certification can be capitalized on over 5 years. You can choose to prepare the entire title progressively over time or have your complementary skills recognized by the VAE.

Click here to discover the evaluation and pedagogical methods of the program  

Performance indicators 2022/2023

• Satisfaction rate: 100%
• Diploma presentation rate: 100%
• Diploma success rate: 100%

Content updated on 11/19/2023

TARGET AUDIENCE AND PRESENTATION

This program is designed for professionals in the field of data science and computer science, or for any scientist and any technical or managerial profile with prior knowledge in linear algebra, probability, statistics and programming (Python), who wish to acquire operational skills in artificial neural networks and deep learning. The program also allows students to discover how to ethically and responsibly deploy deep learning solutions. Finally, this program aims to strengthen and foster collaborations between scientific teams and the business departments of a company.

Although the theory of deep learning dates back to the mid-20th century, its application to solve complex problems from a human point of view dates back only about ten years. It was at this time that the growth in computing capacity was able to respond to the strong computational needs for new algorithms, their optimization and the increase in data flows.

Today, Internet giants (Amazon, Google, Facebook, etc.) use deep learning to segment their users and recommend content that is likely to “click”. But in recent years, deep learning has also become open to everyone thanks to the development of open source, knowledge sharing and the miniaturization of graphics processors, capable of training deep neural network models.

Object detection, medical diagnostics, voice recognition, archive restoration, risk prediction and prevention: an enterprise can now build its business model around deep learning application innovation.

OBJECTIVES

This training will allow you to acquire:

• An in-depth understanding of the elements of a neural network and their relationships
• Concrete experience with current deep learning methods and their application to business cases.
• All the tools and skills to implement deep learning solutions that are business oriented but also socially and ethically responsible.

Contenu

• 

  Module 1 : Introduction aux réseaux de neurones

  Fondamentaux des réseaux de neurones artificiels, leur théorie et leurs applications business.

  Les principaux concepts abordés dans ce module sont : les réseaux de neurones, la propagation et la rétropropagation, l’initialisation, la régularisation et l’optimisation, la sélection des hyperparamètres.

• 

  Module 2 : Méthodes actuelles de l'apprentissage profond

  Exploration des méthodes actuelles de deep learning et des façons de les appliquer dans un contexte business.

  Les principaux concepts abordés dans ce module sont : les réseaux de neurones convolutifs, la vision par ordinateur, les réseaux de neurones récurrents et la mémoire court-terme et long terme.

• 

  Module 3 : Déploiement responsable du deep learning

  Déploiement d’une solution responsable basée sur le deep learning, en se concentrant sur l’interprétabilité et les implications sociales et éthiques.

  Les principaux concepts abordés dans ce module sont : l’interprétabilité, le déploiement de modèle, le flux de travail typique d'un projet d’intelligence artificielle.