Developing and testing methodologies that allow to interpret the predictions of the AI algorithms in terms of transparency, interpretability, and explainability has become today one of the most important open questions in AI.
In this proposal we bring together researchers from different fields with complementary skills, essential to be able to understand the behaviour of the AI algorithms, that will be studied with an interesting set of multidisciplinary use-cases in which explainable AI can play a crucial role and that will be used to quantify strengths and highlight, and possible solve, weakness of the available explainable AI methods in different applicative contexts. One aspect hindering so far substantial progress towards explainability is the fact that several proposed solutions in explainable AI proved to be effective after being tailored to specific applications, and frequently not easily transferred to other domains. In this project, we will test the same array of techniques for explainability to use-cases intentionally chosen to be quite heterogeneous with respect to the types of data, learning tasks, scientific questions.
The proposed use-cases range from High Energy Physics  AI applications, to applied AI in medical imaging, to applied AI for the diagnosis of pulmonary, to tracheal and nasal airways, to machine-learning techniques of explainability used to improve analysis and modelling in neuroscience.
For each use-case, the research project will consist of three phases. In the first part, we will apply state-of-the-art explainability techniques, properly chosen based on the requirements, to the case under consideration. In the second part, shortcomings of the techniques will be identified. Most notably, issues of scalability to high-dimensional and raw data, where noise can be prevalent compared to the signal of interest, will be taken into consideration, as long as the level of certifiability afforded by each algorithm. In the final phase, new algorithmic methodologies adequate to HEP, medical, and neuroscientific use cases will be designed, based on these considerations

Call Topic: Explainable Machine Learning-based Artificial Intelligence (XAI), Call 2019
Start date: (36 months)
Funding support: 890 250 €

Main results

PUBBLICATIONS

  1. Broboana, D., Bratu, A.M., Magos, I., Patrascu, C. and Balan, C., 2022. Kinematics of the viscous filament during the droplet breakup in air. Scientific reports, 81(11), 12(1), pp.1-9. Link  

  2. Spinelli, I., Scardapane, S. and Uncini, A., 2022. A Meta-Learning Approach for Training Explainable Graph Neural Networks, IEEE Transactions on Neural Networks and Learning Systems, early access, pp. 1-9.  Link    arXiv    Code  

  3. Botta, D., Magos, I. and Balan, C., 2021. The Influence of Surface Tension on Radial Wicking in Paper. 10th International Conference on ENERGY and ENVIRONMENT (CIEM), (pp. 1-5). IEEE. Link  

  4. Botta, D., Magos, I. and Balan, C., 2021. Influence of Viscosity on Radial Diffusion of Fluids in Paper Substrates. 2021 12th International Symposium on Advanced Topics in Electrical Engineering (ATEE), (pp. 1-5). IEEE.  Link  

  5. Chiriac, E., Avram, M. and Balan, C., 2021. Transition from Threads to Droplets in a Microchannel for Liquids with No Viscosity Contrast. 10th International Conference on ENERGY and ENVIRONMENT (CIEM), (pp. 1-4). IEEE. Link  

  6. Chiriac, E., Bran, A.M., Voitincu, C. and Balan, C., 2021. Experimental validation of VOF method in microchannel flows. 2021 12th International Symposium on Advanced Topics in Electrical Engineering (ATEE), (pp. 1-4). IEEE.  Link  

  7. Francescato, S., Giagu, S., Riti, F., Russo, G., Sabetta, L. and Tortonesi, F., 2021. Model compression and simplification pipelines for fast deep neural network inference in FPGAs in HEP. The European Physical Journal C, 81(11), pp. 1-10.  Link  

  8. Magos, I. and Bălan, C., 2021. Impact of Solid Hydrophilic Spheres on Fluid Surfaces. 10th International Conference on ENERGY and ENVIRONMENT (CIEM), (pp. 1-4). IEEE.  Link  

  9. Tanase, N.O., Broboana, D. and Balan, C., 2021. Stability of the Lid-Cavity Flow at Low Reynolds Numbers. 10th International Conference on ENERGY and ENVIRONMENT (CIEM), (pp. 1-4). IEEE. Link 

  10. Tanase, N.O., Enache, A., Broboana, D. and Balan, C., 2021. Experimental and Numerical Studies of the Free Surface Flow over the Patterned Weir. 2021 12th International Symposium on Advanced Topics in Electrical Engineering (ATEE), (pp. 1-4). IEEE. Link  

  11. L Maglianella, L Nicoletti,  S Giagu, C Napoli, S Scardapane, Convergent Approaches to AI Explainability for HEP Muonic Particles Pattern Recognition. Comput Softw Big Sci 7, 8 (2023). Link 

  12. A. Verdone, A. Devoto, C. Sebastiani, J. Carmignani, M. D'Onofrio, S. Giagu, S. Scardapane, M. Panella, Enhancing High-Energy Particle Physics Collision Analysis through Graph Data Attribution Techniques, arXiv:2407.14859 [cs.LG]. Link 

  13. T. Torda, A. Ciardiello, S. Gargiulo, G. Grillo, S. Scardapane, C. Voena, S. Giagu, Influence based explainability of brain tumors segmentation in multimodal Magnetic Resonance Imaging, arXiv:2405.12222 [eess.IV]. Link 

  14. A. Biondini, M. D’Onofrio, S. Giagu, C.D. Sebastiani et al. (ATLAS Collaboration), Search for light long-lived neutral particles that decay to collimated pairs of leptons or light hadrons in collisions at ~TeV with the ATLAS detector, J. High Energ. Phys. 2023, 153 (2023). Link

  15. A. Bran, N.O. Tanase, C. Balan, Interface Dynamics and the Influence of Gravity on Droplet Generation in a Y-microchannel. Micromachines 2022, 13, 1941.Link

  16. G. Grillo, Integrating ChatGPT-4: A Novel XAI Interface for Enhanced Clinician Understanding of MRI Image Segmentation Results, IEEEXplore CBMS 2024. Link

 

DISSEMINATION

Talk at Conferences, Seminars, ...

  1. A. Ciardiello, Explainable deep learning inference to decode decision-making processes from multidimensional patterns of neural activities , Conference: 6th BigBrain Workshop, October 25-27, 2022 in Zadar, Croatia.

  2. S. Giagu, Explainable and Interpretable AI in physics and applied physics, Invited Seminar at University of Milano Bicocca, Nov. 28th 2022

  3. S. Giagu, MUCCA Multi-disciplinary Use Cases for Convergent new Approaches to AI explainability, Workshop AI@INFN, Bologna 2-3 May 2022

  4. Ciardiello, From which to why: interpretation map for Explainable Deep Learning based on influence methods, Workshop MSMM 2023, Torino 30-31 May 2023 

  5. S. Scardapane, Advancing Explainable AI: Testing and Enhancing Techniques Across Multidisciplinary Use-Cases, Workshop on Artificial Intelligence and the Uncertainty challenge in Fundamental Physics, Paris, 27.11-1.12, 2023

  6. T. Torda, Tracin in Semantic Segmentation of Tumor Brains in MRI, an Extended Approach, HC@AIxIA 2nd AIxIA Workshop on Artificial Intelligence for Healthcare, Rome 2023

  7. G. Grillo, T. Torda, C. Voena, A. Ciardiello, S. Giagu, Integrating ChatGPT-4: A Novel XAI Interface for Enhanced Clinician Understanding of MRI Image Segmentation Results, IEEE CBMS 2024 26-28.6.2024 Guadalajara, Mexico

  8. J. Carmignani, Integrating Explainable AI in Modern High-Energy Physics (the MUCCA Project), EUCaifCon 2024, 30.4-3.5.2024 (Amsterdam): Link 

 

Presentations at CHIST-ERA Annual Projects Seminar

  1. CHIST-ERA PROJECTS SEMINAR 2021: S. Giagu - April 12-14 (online): link

  2. CHIST-ERA PROJECTS SEMINAR 2022: C. Voena - March 28-30 (online): link

  3. CHIST-ERA PROJECTS SEMINAR 2023: A. Ciardiello - April 4-5, Bratislava: link 

  4. CHIEST-ERA PROJECTS SEMINAR 2024: C.D. Sebastiani - Helsinki, Finland to Stockholm, Sweden and back on April 16-18: link  slides 

 

MUCCA General Meetings

  1. 2021: Mucca Kickoff meeting, Feb. 22, 2021 (online)

  2. 2022: First MUCCA General Meeting, April 11-13, at the Orto Botanico in Rome, agenda

  3. 2023: Second MUCCA General Meeting,  June 12-14, Bucharest

 

Schools & Hackhathons 

  1. From Graph Neural Networks to Explainable AI, comprehending and trusting Machine Learning algorithms, School 20-22.9 2023, SPINE Liverpool and University of Liverpool. Link 

Website: https://sites.google.com/uniroma1.it/mucca-project

Attachment

CHIST-ERA Call 2019 - XAI Topic - MUCCA 2022.pdf 7.09 MB
CHIST-ERA Call 2019 - XAI Topic - MUCCA 2021.pdf 3.68 MB

Publications in Open Access