Marta Moscati

Marta Moscati works at the Institute of Computational Perception at Johannes Kepler University Linz, focusing on advanced recommendation systems and multimodal learning. Her research spans emotion-based music recommendation, privacy-preserving machine learning, and graph neural networks.

Recent work includes:

• Developing multimodal single-branch architectures for cold-start scenarios
• Creating preference obfuscation techniques in implicit feedback systems
• Advancing music emotion recognition with semi-supervised graph networks
• Contributing to the FAME Challenge for multilingual face-voice association

She has published extensively in top AI venues while maintaining technical expertise in both deep learning and theoretical physics, with early work on lepton universality violation.

At JKU, she contributes to:

• Recommendation algorithms development
• Multimodal representation learning research
• Musical affective computing applications
• Privacy-preserving AI frameworks