In the coming weeks, we will present the winning projects of the cascade calls (FAIR cascade funding calls) dedicated to Universities and Research Institutions, promoted by the spokes between late 2023 and early 2024. FAIR’s cascade calls are issued by the ten Spokes based on their respective research objectives, with the aim of attracting an increasingly broad range of organizations and high-quality research projects in the field of artificial intelligence into the ecosystem.
The first projects we introduce will collaborate with Spoke 10, coordinated by the Italian Institute of Technology (affiliates include CNR, INFN, the University of Catania, Leonardo, and STMicroelectronics).
What are cascade calls?
Within the “Call for the creation of extended partnerships among universities, research centers, and companies for the funding of basic research projects” included in the research supply chain measures of the National Recovery and Resilience Plan (PNRR), published by the Ministry of University and Research (MUR) in March 2022, it is предусмотрed that the Spokes of the extended partnerships issue calls to grant funding to external entities for research activities and for the procurement of supplies, goods, and services necessary for their implementation.
The first cascade calls issued by FAIR were dedicated to research projects carried out by public universities, non-public universities legally recognized and accredited by MUR, and public research institutions supervised by MUR, for a total budget of 18 million euros distributed among the ten spokes of the project. A second round of funding, still open, is instead dedicated to cascade calls for micro, small, medium, and large enterprises.
Winners of the Spoke 10 call
Spoke 10 focuses its research activity on bio-socio-cognitive AI: its goal is to design so-called bioinspired artificial intelligence systems, inspired by models of human learning.
These are the winning projects of the call issued by Spoke 10, with a total funding of €2,670,000 distributed across 12 research objectives:
Object ReCognition datA – ORCA
Project Coordinator: Prof. Marco Bertamini, University of Padua
The project focuses on the complexities of object recognition in visual perception. The objective is to collect and analyze a large dataset of electroencephalogram (EEG) recordings, with particular attention to shapes or objects belonging to specific categories.
Adaptive MEta-Learning strategies for concept-drift awareness in distributed Intelligent Systems – AMELIS
Project Coordinator: Prof. Salvatore Gaglio, University of Palermo
The project aims to define new adaptive learning strategies to make artificial intelligence systems more flexible within distributed and heterogeneous perceptual infrastructures. It integrates supervised and unsupervised learning techniques for anomaly detection, ensemble learning methods to overcome the limits of individual algorithms, and meta-learning strategies to flexibly orchestrate intelligent components.
artiFicial And bio-inspIred netwoRked intelliGence foR cOnstrained aUtoNomous Devices – FAIRGROUND
Project Coordinator: Prof. Dario Bruneo, University of Messina
FAIRGROUND aims to reshape the paradigm of intelligent systems by developing new neuromorphic architectures, non-backpropagation learning algorithms, and bio-inspired Spiking Neural Network (SNN) algorithms, enabling broader adoption of AI in constrained devices capable of autonomous learning in real-world environments.
Self-conscious behavior in Embodied AI agents – CAESAR
Project Coordinator: Prof. Antonio Chella, University of Palermo
This project develops AI methods to support robots in generating self-aware behaviors based on models of themselves, their components, and the environment. It also focuses on behaviors that enable robots to explain their motivations and actions.
Visual Attention for New-generation of Vision Transformers – VisAViT
Project Coordinator: Prof. Massimo Tistarelli, University of Sassari
The project leverages knowledge of human visual attention mechanisms to improve vision transformer architectures by introducing new data primitives, reducing training time and data requirements, enhancing self-attention with relevance and saliency, and increasing robustness against adversarial attacks. Face recognition will be used as a benchmark application.
innovAtive human-iN-the-loop-baSed knowledge undERstanding – ANSWER
Project Coordinator: Alessandro Sebastian Podda, University of Cagliari
The project aims to develop an innovative AI method to extract knowledge from multimodal data, creating Knowledge Graphs enhanced through human-in-the-loop pipelines and integrating multiple AI techniques including LLMs and information extraction.
Harmonic Analysis and Optimization in Infinite-Dimensional Statistical Learning – HAOISL
Project Coordinator: Prof. Ernesto De Vito, University of Genoa
The project seeks to build a mathematically grounded understanding of machine learning, integrating high-dimensional probability, optimization, and numerical analysis to design efficient algorithms and support applications in inverse problems and imaging.
Foundations of certifiably lightweight and secure AI systems – CLASH-AI
Project Coordinator: Prof. Lorenzo Rosasco, University of Genoa
CLASH-AI addresses the trade-off between robustness and efficiency in machine learning models, developing indicators to certify security and sustainability and validating results through practical use cases.
LEveraging computer VIsion and Robotics for human-centered AI – LEVIR
Project Coordinator: Prof. Francesca Odone, University of Genoa
LEVIR focuses on AI methods for vision, robotics, and intelligent systems capable of handling complex data and interacting naturally with humans, proposing new approaches in computer vision and adaptive social robotics.
Grounded deep learning models for numerical cognition – GROUNDEEP
Project Coordinator: Prof. Marco Zorzi, University of Padua
The project develops biologically inspired generative architectures to model numerical cognition, comparing grounded and non-grounded models to evaluate improvements in numerical concept representation.
Robust Models for Safe and Secure AI Systems – SAFER
Project Coordinator: Prof. Battista Biggio, University of Cagliari
SAFER develops scalable testing methodologies and robust AI models capable of withstanding realistic adversarial manipulations, integrating them into an MLSecOps framework for secure deployment.