Albert Gatt

Albert Gatt is a researcher at the University of Malta, with extensive contributions to Natural Language Generation (NLG), Vision-and-Language (V&L) models, and evaluation practices in NLP. His work spans multimodal reasoning, data pruning efficiency, and reproducibility challenges in human evaluations.

• Key collaborations include studies on temporal grounding in image sequences (TempVS benchmark) and automated legal violation detection in cookie banners.
• Research highlights include bridging linguistic theory with computational models (e.g., VALSE benchmark for multimodal grounding) and improving generation quality through contrastive learning frameworks.

Scientific awards are not explicitly mentioned in the provided texts. His work emphasizes rigor in automatic metric validation and cross-modal interpretability, particularly in multimodal model attention mechanisms and logical formula minimization for text generation.