PhD-Cooperation schemes & heterogeneous algorithmic approaches F/MOrange

Orange Innovation brings together the research and innovation activities and expertise of the Group's entities and countries. We work every day to ensure that Orange is recognized as an innovative operator by its customers and we create value for the Group and the Brand in each of our projects. With 720 researchers, thousands of marketers, developers, designers and data analysts, it is the expertise of our 6,000 employees that fuels this ambition every day.
Orange Innovation anticipates technological breakthroughs and supports the Group's countries and entities in making the best technological choices to meet the needs of our consumer and business customers.
Within the Innovation department, you will be integrated into a research team at the forefront of innovation and expertise in artificial intelligence and cybersecurity. You will be part of a research ecosystem involved in the optimization of networks and services of the Orange Group. The thesis will take place at Orange Research and at the EURECOM laboratory (SophiaTech). It will be supervised academically by Dr. Raphaël Troncy and industrially by Dr. Lionel Tailhardat.

Your role is to conduct thesis research on the topic of 'Optimizing cybersecurity incident diagnostics in the era of generative AI: cooperation schemes and self-orchestration of heterogeneous algorithmic approaches under time and explainability constraints.'
Global Context and Problem Statement
Modern computer networks and telecommunications services, due to their complexity, pose unprecedented challenges for rapid incident diagnosis. The dispersion of expertise and technical approaches within an organization notably represents an opportunity for malicious actors. Therefore, it is crucial to develop methods and tools for rapid and effective diagnosis in these complex environments.
Scientific Objective - Challenges and Expected Results
To understand how to automatically orchestrate, under time and explainability constraints, various algorithmic solutions for detecting and diagnosing complex anomalies. A set of detection and diagnostic techniques and models is currently available and used by the R&D community and industry. Each model typically has maximum efficiency for a subset of anomalies among all failure modes and undesirable situations that can occur on a network. We hypothesize that a form of collaboration between these techniques and models (agents) can be established to maximize the coverage of these anomalies and their remediation. So the main scientific and technical challenges are :
Multi-model architectures for alarm correlation through synergistic reasoning: (to theorize and design a reference architecture adapted to the application domain to differentiate the coexisting options to date).Explainability & transferability of behaviors and beliefs between models: (to develop an abstraction that allows establishing cooperation between heterogeneous agents while minimizing transaction costs and maximizing reusability between technical ecosystems).Self-organization & diagnostic strategies under time, explainability, and efficiency constraints: (to understand the effectiveness of emerging diagnostic strategies, their conditions of emergence, and their replicability in other contexts).
To address these research axes, the PhD student will be required to:Conduct a state-of-the-art review on the design paradigms of multi-agent systems, with a perspective for NetOps/SecOps.Develop a prototype of a multi-agent system for collective causal inference.Study the strategies and conditions for engagement, invocation, and cooperation between models (agents).

Education
You hold a professional or research master's degree, or you are a graduate of an engineering school in computer science or applied mathematics, preferably with a specialization in one or more areas of artificial intelligence.
Experience from an internship or project related to one of the following activities would be an asset for this position:
Design and deployment of a supervision/decision support system.
Design and use of machine learning algorithms and/or combinatorial optimization.
Implementation of Semantic Web technologies.
Implementation of multi-agent systems.
Management, control, and auditing of telecommunications networks and services.
Skills and Personal Qualities
Knowledge of machine learning and/or automated reasoning, with implementation of algorithms.
Knowledge of applied mathematics, computation, and simulation, particularly with a focus on optimal control, and dynamic and distributed systems.
Proficiency in Semantic Web technologies (RDF/RDFS, SPARQL).
Software development skills (Python, Java, C++, Prolog, ASP, LISP).
Excellent writing and presentation skills in French and English.
Personal qualities: problem-solving, autonomy, proactivity, rigor, analytical skills, and initiative.

Intéressement, participation, mutuelle, restaurant d'entreprise, participation frais de transport, activités sociales et culturelles CSE