fr
en
Implement algorithms and development techniques for AI
Your AI ideas need to translate into processing steps and code. Structure your programming practice to manipulate data and implement algorithms. Develop an approach that connects technical logic, experimentation and verification of results.
- Duration
- 2 days 14 hours
- Code
- IA046FR Code
Presentation
Developing artificial intelligence involves more than calling APIs: it requires a detailed understanding of the underlying algorithms. This intensive 2-day course is designed specifically for developers and software engineers seeking strong technical expertise in machine learning and deep learning.
The programme goes beyond theory and takes you into the code. You will learn to prepare data rigorously, a crucial step in any AI project, before implementing supervised and unsupervised learning algorithms, including regression, SVM and random forests. You will then progress to neural networks and deep learning, using leading frameworks such as TensorFlow and Keras.
Firmly practical, the course alternates applied mathematical concepts with coding workshops. You will design, train and optimise your own models on real datasets, such as Titanic and MNIST, enabling you to choose the right algorithmic architecture for each problem.
Objectives
By the end of this course, you will be able to:
- analyse and select machine learning or deep learning algorithms suited to the requirement;
- apply data preparation techniques, including cleaning, encoding and normalisation;
- implement high-performing predictive models with Python libraries such as Scikit-learn and TensorFlow;
- design neural network architectures, including CNNs and RNNs, for complex data;
- evaluate and optimise model performance through hyperparameter tuning and cross-validation.
Program
Day 1: Mastering algorithmic fundamentals and classical machine learning
- Key concepts: supervised versus unsupervised learning, classification and regression.
- Data preparation: cleaning, encoding, normalisation and handling imbalanced data.
- Core algorithms: linear/logistic regression, decision trees, random forests, KNN and SVM.
Hands-on exercises
- Implement a complete classification model with Scikit-learn on a real dataset, such as Iris or Titanic.
Day 2: Exploring deep learning and advanced architectures
- Deep learning fundamentals: artificial neurons, activation functions and backpropagation.
- Getting started with frameworks: comparing TensorFlow and PyTorch, and using Jupyter/Colab.
- Advanced architectures: multilayer perceptrons (MLPs), convolutional networks (CNNs) for images and recurrent networks (RNNs) for time series.
- Optimisation: analysing ROC curves, confusion matrices and cross-validation.
Hands-on exercises
- Create and train a neural network with TensorFlow/Keras for image recognition using MNIST or CIFAR-10.
Audience
This course is intended for technical professionals seeking to specialise, including:
- Python developers seeking to move into AI engineering or data science roles;
- software engineers and data engineers who need to understand the algorithmic black box to integrate or optimise models more effectively;
- computer science or data science students seeking to strengthen theoretical knowledge through intensive practice;
- technical project managers in R&D or innovation seeking to master technical concepts to lead expert teams more effectively.
Prerequisites
The following prerequisites apply:
- Technical experience: a sound foundation in Python programming.
- Theoretical knowledge: general mathematical knowledge, including statistics and linear algebra.
- Advantage: initial experience manipulating data will support learning.
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 depth: go beyond an introduction to understand algorithms' internal mechanisms, including backpropagation and activation.
- Modern stack: use industry-standard tools, including Scikit-learn, TensorFlow, Keras and PyTorch.
- Real projects: validate your skills on reference datasets, including Titanic and MNIST, widely used by data scientists.
- Complete perspective: cover the full spectrum from classical machine learning, or shallow learning, to deep neural networks.
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 Python, TensorFlow, PyTorch and Scikit-learn, belong to their respective owners. Their mention for educational purposes does not constitute an endorsement or partnership.