fr
en
Deep Learning fundamentals
Deep learning becomes usable when you understand the choices that influence its results. Connect architectures, data and training to analyse model behaviour. Develop a framework for designing experiments and assessing their performance rigorously.
- Duration
- 1 day 7 hours
- Code
- IA035FR Code
Presentation
At the heart of the artificial intelligence revolution, deep learning enables major advances in image recognition, language processing and predictive analytics. This one-day course is designed to demystify technologies often seen as complex and opaque. It is intended for professionals who want to understand how machines learn from large volumes of data.
The programme explores the inner workings of artificial neural networks. You will discover the fundamental differences from traditional machine learning and examine key architectures, such as convolutional networks (CNNs) for images and recurrent networks (RNNs) for sequences. The teaching approach focuses on using accessible tools such as Keras and TensorFlow to reinforce theoretical concepts.
Through hands-on workshops, you will move from theory to practice by building and evaluating your first simple models. You will leave with a clear understanding of deep learning's actual capabilities, enabling you to identify relevant use cases in your sector and communicate effectively with technical experts.
Objectives
By the end of this course, you will be able to:
- distinguish the principles of deep learning and its place within AI;
- identify the main neural network architectures, including MLP, CNN and RNN, and their applications;
- describe the key stages of building and training a neural model;
- interpret results and performance indicators to evaluate a model;
- use leading libraries such as TensorFlow and Keras for simple use cases.
Program
Module 1: Introduction to artificial neural networks
- The definition and history of deep learning, and its differences from machine learning.
- How an artificial neuron, activation functions and backpropagation work.
- Multilayer perceptron (MLP) architecture and application areas, including vision and NLP.
Hands-on exercises
- Map potential deep learning use cases in your sector.
- Build a simple neural network using the Keras library.
Module 2: Exploring advanced architectures
- Convolutional neural networks (CNNs) specialised in image processing and classification.
- Recurrent networks (RNNs and LSTMs) suited to time-series analysis.
- Transformer fundamentals: the technology behind modern generative AI.
Hands-on exercises
- Explore and use a pre-trained CNN model for image classification.
Module 3: Evaluating and optimising models
- Analysing performance metrics, including accuracy and loss, and the confusion matrix.
- Understanding overfitting and regularisation techniques.
- Interpretability challenges in understanding black-box decisions.
Hands-on exercises
- Analyse a model's results and adjust its parameters to improve performance.
Audience
This course is intended for technical and business professionals seeking to build their expertise, including:
- product owners and AI project managers responsible for leading complex initiatives;
- developers and data science beginners seeking to specialise in deep learning;
- innovation managers assessing the technology's potential for their business.
Prerequisites
The following prerequisites apply:
- Professional experience: familiarity with basic machine learning or statistical concepts.
- Basic knowledge: elementary Python programming knowledge is recommended for the workshops.
Teaching and assessment methods
- Initial skills assessment
- Training materials provided to participants
- Continuous assessment throughout the course
- End-of-course feedback questionnaire
- Combination of theory and practical application
- Attendance records
- Post-course follow-up evaluation
- Practical exercises
Course highlights
- Technical immersion: discover industry-standard libraries, including TensorFlow and Keras, through practical, guided exercises.
- Structured understanding: grasp the logic behind different architectures, including CNNs and RNNs, to identify which technology suits each problem.
- Pragmatic approach: move beyond theory to build, train and, above all, evaluate the relevance of your own models.
- Accelerated format: acquire the essential understanding needed to navigate deep learning in a single day.
Dates and sessions
Choose the date and delivery format that suit you.
No upcoming sessions are currently available.
Any brand names and logos mentioned in this course description, such as TensorFlow, Keras and PyTorch, belong to their respective owners. Their mention for educational purposes does not constitute an endorsement or partnership.