"Implementing Data Science and Deep Learning for Artificial Intelligence"

Training Duration: 98 hours
Dates: See the calendar below

Location: Campus Pierre et Marie Curie – Paris (Jussieu)
Fee: 8 000€ + 254€ of university inscription fees
Format: In-person
CPF: Eligible (The CPF, or Compte personnel de formation, is a French government program that allows individuals to finance professional training. Stating it is "Eligible" means participants can use their CPF funds to pay for the course.)
Qualification: Diploma course

Contact: ingenierie-fc@sorbonne-universite.fr

 

Training Objectives  

The Machine Learning and Artificial Intelligence certification aims to complement and enrich the skills of employees in the big data sector. The goal is to enable them to perform predictive analysis that addresses business challenges.

The unique aspect of this training is its equal focus on computer science and mathematics, while maintaining a strong professional orientation.

This program is designed to allow graduates to advance their careers quickly in the data analysis and information technology sectors. By closely combining a technical foundation in computer science, mathematics, and business knowledge, it will give participants the ability to model, code, and analyze the results and implications of a data science approach. It will also enable graduates to work in the consulting sector, where there is a growing demand for data experts, or to develop new industrial approaches within a company that wants to leverage its data.

This certification is based on one of the 34 plans of the "Nouvelle France Industrielle" (New Industrial France) dedicated to Big Data, which has been transformed into the Data Economy solution. Its 2020 goal was to create and consolidate 137,000 jobs.

 

Learning Outcomes 

Upon completion of the certification, the trainee will be able to:

• Master the mathematical and computer science fundamentals for data processing.
• Link machine learning tasks to specific business problems.
• Identify the algorithmic needs to address these tasks.
• Develop a Deep Learning architecture.
• Understand the societal challenges of artificial intelligence.
• Develop a Big Data architecture.
• Successfully lead a data project.

 

Training Support 

The training includes personalized follow-up, structured around group sessions led by one or more supervising instructors.

 

Target Audience & Prerequisites

This training is designed for a diverse range of professionals who share a strong motivation to acquire skills in data science.

The course is intended for technicians and managers with a solid background in mathematics and/or computer science. They should be looking to develop expertise in machine learning, deep learning, artificial intelligence, or big data processing.

 

PROGRAM

Module 1: Mathematics and Computer Science Tools

This module aims to bring students up to speed on key skills in mathematics and computer science. Topics covered include basic algorithms, mastering the Python language and its data processing libraries (numpy, pandas, matplotlib, bokeh, etc.), and managing input/output. It also covers analysis (integration), linear algebra (matrix products and inversion, eigenvalues, and singular values), and basic probabilities and statistics (probability space, sampling, statistical models, inference).

Format: 2 days of lectures/hands-on workshops.

 

Module 2: Supervised Statistical Learning

This module provides an overview of today's statistical learning landscape. It addresses the major problems in the field and presents the key advances of the last ten years, with a focus on supervised models.

General framework of supervised learning
Naive Bayesian classifier
Overfitting and cross-validation
Trees: CART, random forests, and boosting
Empirical risk minimization
Logistic regression and support vector machines
Perceptron, neural networks
Regularized approaches
Processing uncertain and incomplete data

Format: 2 days of lectures/hands-on workshops.

 

Module 3: Unsupervised and Deep Learning

This course focuses on deep neural network models and the associated computing tools.

Gradient descent algorithms, back-propagation, and their variants
Convolutional neural networks and recurrent networks
Tools: Declarative platforms (TensorFlow) and algorithmic frameworks (PyTorch)

Format: 2 days of lectures/hands-on workshops.

 

Module 4: Deep Learning and Neural Networks for Images

This module's objective is to study deep neural networks for automatic image recognition and the classic algorithms used in unsupervised learning.

Topics include:

• SIFT descriptors and neural network architectures applied to images (e.g., convolutional networks).
• Clustering: Hierarchical classification, k-means and its variants, Gaussian mixture models, and spectral methods.
• Dimensionality reduction and visualization: Singular value decomposition, PCA and its variants, MCA, MDS, ISOMAP, and t-SNE.

Format: 2 days of lectures/hands-on workshops.

 

Module 5: Neural Networks for Language Processing and Data Environment

This module aims to study neural networks applied to text data and to understand the broader context of industrial approaches related to Big Data. It will particularly make participants aware of regulation and ethics, and emphasize the integration of data science methodologies into a professional environment.

Topics include:

• Word representations and neural network architectures applied to text (recurrent, attention, and "adversarial" networks).
• Data governance and ethics.
• Cybersecurity and business analytics.
• Communication and "data storytelling."

Format: 2 days of lectures/hands-on workshops.

 

Module 6: Cloud Computing and Big Data

The objective of this module is to train students on tools for managing big data. This course is focused on computer tools and infrastructure, covering the following topics:

• SQL and NoSQL databases
• Hadoop, Spark, MapReduce
• Cloud Computing and virtualization
• Pig, Hive, and SPARQL

Format: 2 days of lectures/hands-on workshops.

 

Module 8: Data Project

The goal of this module is to supervise students as they work on data processing projects. The project topics and the data to be analyzed will be proposed by the students themselves. A special emphasis will be placed on the presentation (data storytelling) of the results and the models developed.

Format: Personalized follow-up, structured around group sessions with one or more supervising instructors.

 

Assessment Methods
 

The certification's evaluation is based on two main components:

Continuous Assessment: This involves ongoing knowledge checks within each of the 8 modules.
Final Project: Students must complete a project and defend it before a jury.
 
 

Certification Requirements

To receive the certification, a student must meet the following criteria:

Knowledge Evaluation: Achieve a score of 10/20 or higher on the continuous assessments, which are conducted through quizzes or graded exercises.
Project Evaluation: Achieve a score of 10/20 or higher on the final project defense.
The certification is a Sorbonne University diploma, awarded by the training jury.

 

Career Opportunities

This training allows individuals to secure their professional path by providing them with the necessary skills to help companies address the challenges in their sector and adapt to associated technological developments.