Data-agnostic Learning for Multimodal Integration and Generation – DL-MIG

Funder: Project funded under the National Recovery and Resilience Plan (NRRP), Mission 4 Component 2 Investment 1.3 – Cascade Call related to the Program “Future Artificial Intelligence Research (FAIR)” funded by the European Union – NextGenerationEU;

Award Number: Cascade Call related to the Program “Future Artificial Intelligence Research (FAIR)” PE00000013, CUP J33C22002830006;

Start date: 01/04/2024

End date: 31/03/2025

Total NRRP funding for the cascade call: € 1.654.532,00 

Total NRRP funding for Unipr: € 197.657,00

Objectives

FAIR intends to contribute significantly to the objectives set forth in the Italian Strategic Programme on Artificial Intelligence:

• Advance frontier research in AI
• Reduce Artificial Intelligence research fragmentation, foster critical mass and inclusion.
• Create human-centred, robust, trustworthy and sustainable AI
• Foster AI-based innovation and development of AI technology
• Create, retain, and attract AI talent in Italy.
• Ensuring the long-term sustainability of the FAIR Hub

Following the Italian National Strategy in Artificial Intelligence, the FAIR project combines an inclusive and interdisciplinary approach with the valorisation of the excellence widespread in the Italian scientific community in Artificial Intelligence.

Activities

The DL-MIG project seeks to investigate generative AI solutions to seamlessly integrate various types of multimedia data into a common framework, aiming to surpass limitations of existing models which are either limited to data pairs or require ad-hoc modules with a universal framework capable of handling multiple data modalities. The project’s goal is to investigate solutions for designing a flexible system that can work with any mix of data types, focusing on models that can generate new data from different sources. The aim is to improve the current technology with a framework that can easily combine various data forms for controllable multi-modal generation.

The University of Parma collaborates within Spoke 8, titled “Pervasive AI”, within the Work Package (WP) 5, titled “Multimodal Learning for Language, Audio, and Image Integration”

• Department of Engineering and Architecture

• Università degli Studi di Parma (UNIPR)

The main outcome of the project will consist of new solutions for developing a data-agnostic deep generative framework, capable of surpassing current limitations of the state-of-the-art. Research findings and architectural prototypes will be disseminated in the scientific community mostly by means of scientific publications and by publicly releasing the code for research purposes.

https://fondazione-fair.it/