PhD in Video Content Security using Deep Learning Coding Architectures University of Paris-Saclay in France

PhD Opportunity: Video Content Security in Deep Learning Coding Architectures

The University of Paris-Saclay, in collaboration with CentraleSupélec, CRIStAL (Lille), and LIRMM (Montpellier), is offering a fully funded PhD position focused on the security of video content within deep learning-based coding architectures. The project is supervised by Frederic Dufaux (Directeur de Recherche CNRS), Giuseppe Valenzise (Co-Director, CNRS), Vincent Itier (IMT Nord Europe - CRIStAL, Lille), and William Puech (LIRMM, Montpellier).

Research Areas: The thesis will address the confidentiality and integrity of video content in the context of emerging deep learning video coding standards. Traditional video compression algorithms (e.g., H.264/AVC, HEVC, VVC) rely on hybrid architectures and manual design, but recent advances in deep learning have enabled end-to-end optimized coding architectures with competitive performance. However, the security aspects of these new architectures remain largely unexplored.

Project Objectives: The research will focus on two main aspects: (1) preserving video confidentiality by studying encryption or obfuscation of latent space variables after quantization but before entropy encoding, and (2) verifying content integrity using hash functions and digital signatures in the latent space. The approach aims to enable selective encryption (e.g., blurring faces in surveillance videos) and robust integrity verification, addressing new challenges in multimedia data protection.

Collaboration and Funding: The project is part of the PEPR cybersécurité initiative and involves collaboration between leading French research laboratories. The doctoral researcher will have access to advanced computing resources and is expected to contribute to international publications and conferences.

Eligibility: Candidates should have an engineering or Master's degree (or equivalent) with strong knowledge in image/video processing, security, and deep learning. Upper-intermediate English proficiency is required. The position is open to both French and international applicants.

Application Details: The starting date is October 1, 2026. The application deadline is May 31, 2026. For more information and to apply, visit the official ADUM page or contact the supervisors via LinkedIn.

Keywords: video coding, deep learning, security, confidentiality, integrity, image processing, encryption, digital signature, latent space.