Enseignant chercheur
Publié le
Thématiques de recherche : I conduct research in Multimodal Artificial Intelligence (AI), Natural Language Processing (NLP), and Computer Vision (CV), with a particular focus on document analysis and understanding through multimodal learning. My work aims to enhance document comprehension using advanced approaches such as classification, information extraction, summarization, and question-answering.
I am also interested in the ethical and security challenges of large-scale AI models (LLMs and VLMs), integrating solutions to ensure fairness, transparency, and data privacy. My research includes scene text detection and recognition in complex environments, multilingual information extraction, and identity verification.
Additionally, I develop federated learning and blockchain-based methods to ensure secure document processing. The objective of this work is to create robust, explainable, and scalable AI systems, tailored to both industrial and academic needs.
Points forts des activités de recherche : Dr. Bakkali has made significant contributions to the field of Multimodal Artificial Intelligence (AI), particularly in document understanding by integrating Natural Language Processing (NLP) and Computer Vision (CV). His research focuses on cross-modal learning, enabling AI systems to effectively process and analyze both textual and visual information. He has developed advanced methods for document classification, retrieval, and summarization, leveraging contrastive learning and adaptive inference techniques to improve the efficiency and accuracy of AI models.
A key aspect of his work is domain-adaptive learning, where he explores strategies to enhance AI performance across diverse document types and languages, including low-resource settings. His contributions also extend to secure and privacy-preserving AI, incorporating blockchain-based solutions and federated learning to address trust, transparency, and data protection in AI applications.
Furthermore, Dr. Bakkali’s research addresses ethical challenges in AI, focusing on bias mitigation, fairness, and explainability in large-scale models. His work contributes to the development of scalable and robust multimodal AI systems that can be applied to real-world scenarios, ensuring both performance and responsible AI deployment.
