Federico D'Asaro is a researcher and PhD student at Politecnico di Torino, affiliated with the Department of Control and Computer Science (DAUIN) and the Computer Graphics & Vision Group (CGVG). He works as an external lecturer and teaching assistant for the Applied Data Science Project course in the Data Science and Engineering program. His research focuses on Vision-Language Models (VLMs), particularly addressing the Modality Gap in multimodal feature spaces and their applications in downstream tasks like semantic segmentation and speech emotion recognition. Education: Master's degree in Data Science and Engineering (2021), currently pursuing PhD in Computer and Systems Engineering (39th cycle, 2023-2026). His research at the intersection of Natural Language Processing and Computer Vision investigates how reducing the Modality Gap improves crossmodal performance. Recent work applies Large Speech Models (LSMs) to cross-lingual emotion recognition and non-verbal vocalization tasks. He collaborates with researchers like Andrea Bottino, Giuseppe Rizzo, and Juan José Márquez Villacis on projects involving multimodal deep learning and feature extraction. The trends in his publications highlight expertise in multimodal learning (Vision-Language Models, speech-text alignment), deep learning for segmentation and emotion recognition, and crossmodal adaptation in speech processing. His 2025 work focuses on contrastive alignment and non-verbal vocalization, while 2024 studies explore transfer learning of speech models across languages. Teaching Contributions External lecturer for Applied Data Science Project (2025/26) Course collaborator for Applied Data Science Project (2024/25) Federico is part of the Computer Graphics & Vision Group (CGVG) , contributing to interdisciplinary projects that bridge Computer Vision , Natural Language Processing , and Speech Emotion Recognition . His work emphasizes practical applications of multimodal models in real-world scenarios.





