CASPER · Cross-Modal Alignment for Semantic Processing and Effective Representation

Lines connect the coordinator with its partners.

Project objective

Cross-Modal Alignment for Semantic Processing and Effective Representation (CASPER) project is pioneering a novel framework that synergizes cognitive neuroscience with advanced artificial intelligence methodologies. By employing Graph Neural Networks (GNNs) and Large Language Models (LLMs), CASPER enables seamless integration and representation of diverse modalities, such as text, audio, speech, images, and neural signals, including fMRI, EEG, and MEG. The project’s core aim is to push the frontier of multimodal learning and neural representation decoding, facilitating complex tasks like speech reconstruction, cross-modal content generation, and zero-shot learning. Through the alignment of AI systems with neural activity patterns, CASPER strives to enhance our understanding of human cognition while unlocking groundbreaking applications in neural decoding, assistive technologies, and conversational AI. This transformative approach is poised to redefine the standards of cross-modal representation and neural signal interpretation, marking a significant advancement in the field.

Original text from CORDIS.

Participants

• UNIVERSITA CA' FOSCARI VENEZIA · VeneziaCoordinatorItaly
• ECOLE POLYTECHNIQUE FEDERALE DE LAUSANNE · LausanneSwitzerland
• UNIVERSITE DE GENEVE · GeneveSwitzerland

Links

• View on CORDIS
• DOI: 10.3030/101205348
• https://ec.europa.eu/research/participants/documents/downloadPublic?documentIds=080166e52aaa3082&appId=PPGMS