(M/F) Postdoctoral Position : Data-efficient detection of AI-generated images and videos using multimodal foundation models

Descrizione

(M/F) Postdoctoral Position : Data-efficient detection of AI-generated images and videos using multimodal foundation models Sintesi in italiano (traduzione automatica): L'organizzazione GIPSA-lab, un'unità di ricerca congiunta del CNRS e dell'Università di Grenoble, cerca un ricercatore post-dottorato per un progetto sulla rilevazione di immagini e video generati da intelligenza artificiale. Il candidato si occuperà di sviluppare strumenti di analisi forense in grado di distinguere contenuti reali da quelli sintetici, utilizzando modelli multimodali e metodi di apprendimento non supervisionato o debolmente supervisionato. Le mansioni principali includono la revisione dello stato dell'arte, la validazione dei metodi sviluppati e la pubblicazione dei risultati. È richiesta una laurea di dottorato in un campo pertinente, come Informatica o Ingegneria. La posizione è situata a Grenoble, Francia, e il lavoro può essere svolto in francese o inglese. With rapid advances in generative AI (Artificial Intelligence), it is now very easy to create high-quality synthetic images and videos. Such AI-generated content can be exploited for malicious purposes involving misinformation or disinformation. Consequently, it is crucial to develop reliable forensic analysis tools capable of distinguishing real images and videos from those that are artificially generated. Ideally, these tools should be data-efficient, i.e., requiring little to no training data, and demonstrate strong generalization capabilities when applied to unseen data. In this postdoctoral project, we will address this complex yet essential challenge by leveraging powerful multimodal foundation models combined with suitable unsupervised or weakly-supervised learning methods, applicable to both images and videos. - Conduct a brief state-of-the-art review covering both data-efficient multimedia forensics and multimodal foundation models. - Lead the primary task, in collaboration with the supervisor, of using multimodal foundation models to detect AI-generated images. The focus will be on training-free or weakly-supervised methods that leverage the powerful representations learned by foundation models. - Validate the developed methods using standard benchmarks, specifically evaluating their ability to generalize across a wide range of AI-generated data. - Write papers summarizing the developed methods and submit them to appropriate conferences and journals. - As a secondary task, extend and validate the developed methods for the detection of AI-generated videos. POSITION CONTEXT [Présentation du laboratoire : sa localisation, ses effectifs, ses thématiques de recherche GIPSA-lab is a joint research unit of the CNRS (French National Centre for Scientific Research), Grenoble-INP (Grenoble Institute of Technology), and the University of Grenoble, affiliated with Inria (French National Institute for Research in Digital Science and Technology) and the Grenoble Observatory of Earth Sciences. With 350 people, including approximately 150 doctoral students, GIPSA-lab is a multidisciplinary research unit conducting both fundamental and applied research on signals and complex systems. GIPSA-lab develops projects in the strategic fields of energy, the environment, communication, intelligent systems, life and health technologies, and language engineering. Through its research activities, GIPSA-lab maintains a constant link with the economic environment thanks to strong partnerships with businesses. GIPSA-lab staff are involved in teaching and training at various universities and engineering schools in the Grenoble metropolitan area (Université Grenoble Alpes). GIPSA-lab is internationally recognized for its research in Automation and System Safety, Data Science (Information and Signal Processing), Speech, and Cognition. The unit conducts its research through 11 teams organized into 4 research departments: - Automation and System Safety - Data Science - Geometry, Learning, Information and Algorithms - Speech and Cognition The GIPSA-lab comprises 150 permanent staff and approximately 250 non-permanent staff (doctoral students, post-doctoral researchers, visiting researchers, master's students, etc.) The proposed project will be conducted within the ACTIV (Learning, Classification and Processing of Images and Videos, https://www.gipsa-lab.grenoble-inp.fr/en/team/activ ) team of the Data Science Department at GIPSA-lab. The ACTIV research team and the project supervisor have over 15 years of research experience in the fields of digital forensics and applied machine learning. The working language may be French or English. Annuncio in inglese. Fonte: Euraxess (Commissione europea).