Objective :
The goal of this training course is to enable using deep learning models on various data (image, sound, text, ...).
Target audience :
This course is intended for neophytes in Artificial Intelligence (AI) wishing to explore deep learning techniques for exploiting their data.
Course content :
This course addresses deep learning in a pragmatic way by defining its place in the artificial intelligence ecosystem and then explaining the key concepts regarding the model, the training itself and finally the exploitation of results. Different model architectures are presented in order to acquire a broad view of deep learning. Hands-on exercises as well as demonstrations will promote integration of the methodological concepts in order to develop an enlightened practice and to become familiar with the presented architectures. The last afternoon is dedicated to a workshop session for the participants that need help or support in their personal project. For people that want to attend the workshop, relative information about the project/needs will be asked.
Duration :
2.5 days of training + 0.5 day for workshop
Outline :
1. Neural networks:
- Context, definitions and history
- Fundamentals of deep learning
- Practical work
2. Methodology:
- Data management
- Training and evaluation of a model
- Practical work
3. Reference architectures:
- Convolutional Neural Networks (CNN)
- Practical application
- Recurrent Neural Networks (RNN)
- Transformers with a demonstration
- Graph Neural Networks (GNN)
- Deep reinforcement learning with a demonstration
Prerequisites :
As this is an introductory course, there are no mathematical or AI prerequisites. However, a basic knowledge of Python or another language is desirable for the smooth running of the practical sessions.
