Artificial intelligence and data science

Artificial intelligence (AI) and data science are profoundly transforming the way we understand, model and interact with the world, through algorithms capable of learning from data, optimizing complex decisions, or automatically detecting patterns or anomalies in massive datasets. Today, they are present in fields as varied as health, the environment, the physical sciences and chemistry, the humanities and social sciences, biology, finance, industry and digital technologies.

Artificial intelligence (AI) and data science are built on a solid interdisciplinary foundation. Mathematics andcomputer science provide the fundamental tools for modeling phenomena and analyzing algorithms.Optimization is involved in the search for efficient solutions to complex, often constrained, problems. Probability and statistics are used to model uncertainty, extract useful information from partial or noisy data, and evaluate the performance of AI methods. On the IT side, machinelearning is at the heart of modern AI: it involves methods capable of modeling the underlying phenomenon and automatically improving their performance on the basis of data. Automatic language processing (ALP) aims to enable machines to understand and generate human language. Databases,data mining andknowledge extraction focus on the structuring, management and efficient exploration of large volumes of information. Finally, signal and image processing also play a vital role, providing methods for representing, analyzing, compressing, restoring or automatically extracting information from visual, sound or temporal data, which is often noisy or incomplete.

Alongside learning approaches, AI has historically developed through so-calledsymbolic artificial intelligence approaches, based on explicit representations of knowledge(logical rules, ontologies, deductive reasoning). These methods enable fine-tuned interpretability and formal control of system behavior. Today, there is growing interest in so-called hybrid approaches, which combine symbolic models and machine learning techniques. They aim to take advantage of both the power of data and the richness of logical representations to build more robust, explainable and adaptive systems.

The Faculty of Science draws on its recognized expertise in these fields to offer high-level training courses, supported by dynamic, interdisciplinary research teams. This synergy between teaching and research enables us to train students capable of meeting the scientific, technological and ethical challenges raised by the development of AI and data science.

Research

At the Computer Science and Systems Laboratory (LIS)

The aim of the Data Science cluster is to bring together researchers working on issues centered on data, from a computational point of view, whether in terms of representation, manipulation or processing. The cluster's strength lies in the fact that it involves some fifty researchers covering a broad spectrum, from theory (machine learning, deep learning, data mining, linguistic computing) to applications (information retrieval, content recommendation, automatic language processing, vision, bioacoustics, digital humanities, information systems, human-machine communication). The cluster is structured around 4 main areas of activity:

• Artificial intelligence and learning
• Language and information retrieval
• Multimodality and interaction
• Data management and mining for knowledge extraction

The Signal-Image cluster aims to develop research in image and signal processing, analysis and modeling. This theoretical and/or applied research is closely linked to applications with major societal implications. The Signal-Image cluster comprises two teams, I&M and SIIM, whose research themes are both similar and complementary. In fact, both teams are heavily involved in image and modeling issues.

The Computing division contributes to the development of theoretical and practical computer science. Its research activities cover algorithmics, logic, computational models, artificial intelligence and the geometry/topology of computation. The cluster's research activities in the field of artificial intelligence focus on its formal and algorithmic aspects, in particular in the COALA and LIRICA teams. On the one hand, they concern knowledge representation and reasoning modeling, automatic demonstration, ranging from SAT solvers to proof systems designed for more complex logics, and issues linked to constraint-based reasoning in terms of CSP models; and on the other hand, machine learning.

The System Analysis and Control cluster develops research into the analysis, estimation, control and diagnosis of systems. AI-related activities include estimation and decision support, knowledge acquisition and representation, and reasoning for model building,

At theInstitut de Mathématiques de Marseille (I2M)

The ALEA group is divided into four interrelated teams: probability, statistics, signal and image processing, and mathematics for biology. In the field of artificial intelligence and data science, researchers in this group work on

• statistical learning (Bayesian and variational methods, optimization)
• probabilistic and statistical modeling
• mathematical signal and image processing
• multiresolution tools (wavelets, etc.), time-frequency analysis and data representation
• machine learning and statistics for actuaries
• the application of statistics and machine learning in many fields (industrial or scientific), and interdisciplinary collaborations.

At theInstitut Laënnec

As one of Aix-Marseille Université's research institutes, the Institut Laënnec strengthens the link between research and training, promotion, internationalization and interdisciplinarity, around artificial intelligence and digital health. The Faculty of Science is a key member of this institute, whose ambition is to bring the power of Artificial Intelligence and digital sciences to the patient's bedside.

In other research units

Coming soon