This page was translated by an AI (LLM) with a cursory human check and is awaiting full review.
Deep Learning Architectures (ArchDL)
Manager : Kamel Guerda
Instructors : Members of the IDRIS support team
The objective of this training is to familiarise participants with a diversity of neural network architectures adapted to various data types. It presents different model architectures to offer an extended perspective on Deep Learning.
Objectives
- Understand the principles and functionalities of advanced neural network architectures, such as CNNs, RNNs, Transformers, GNNs and diffusion models.
- Acquire practical skills by implementing these architectures during practical work.
- Apply techniques to adapt these architectures to different types of data, exploring their applications on various data types such as images, sound, text or graphs.
- Target audience
- Prerequisites
- Duration and practical info
- Course content
- Course materials
- Upcoming sessions
Target audience
This training is designed for people having had a first contact with deep learning and neural networks, whether through self-taught, professional, or academic experience. It is particularly suitable for those who have taken the IPDL training, which provides the necessary foundations to grasp the more advanced concepts of ArchDL. IPDL and ArchDL are systematically held back-to-back allowing registration for both courses.
The targeted profiles include engineers, researchers, developers, technicians, PhD students, and project managers who wish to deepen their understanding of deep learning architectures.
Prerequisites
Participants must have an understanding of the basics in algebra and statistics, as well as knowledge of Python syntax, which are necessary for the practical sessions.
Furthermore, a first contact with deep learning is essential, whether through thePractical Introduction to Deep Learning (IPDL) training or another equivalent initial training, such as the first sequences of FIDLE. This prior experience in deep learning is crucial for effective assimilation of the advanced concepts covered in ArchDL.
Duration and practical info
This training lasts 2 days. Welcome is from 09:00 with classes starting at 9:30. Supervision is provided until 18:00 with an average finish at 17:30 (varying by group depending on practicals).
It takes place exclusively in-person at the premises ofIDRIS in Orsay (91).
Attendance
Minimum : 10 people ;
Maximum : 18 people.
Course content
This training presents different model architectures to offer an extended perspective on deep learning. Participants will explore key architectures such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Transformers, and others.
Practical exercises in Python with PyTorch, as well as demonstrations on the Jean Zay supercomputer, will allow these concepts to be put into practice.
Plan
-
Convolutional Neural Network (CNN)
-
Recurrent Neural Network (RNN)
-
Transformers
-
Graph Neural Network (GNN)
-
Diffusion models
For an efficient execution of the practical parts, these will take place on the Jean Zay supercomputer. A workstation with access to the IDRIS supercomputer is provided to the learners. Experience in using a supercomputer, as well as prior access to it, are not required.
Course materials
All course materials, including slides, notes, and practical exercises, are provided under the following license: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0). For more details on the license, please consult this page.
To view the dates of the upcoming sessions for this training, go to the following page:
Registration
CNRS/French university staff | External participants |
Are you a member of CNRS or a French university? Your registration is free via our server. | Our training is aimed at all professionals from companies, public bodies and individuals. |
Consider checking out thePractical Introduction to Deep Learning (IPDL) training and registering if you do not have this prerequisite.