ELEONORA GRASSUCCI

Dottoressa di ricerca

ciclo: XXXV


Titolo della tesi: Hypercomplex Generative Models: AI Beyond Clifford Algebra

Real-world data is complex and often multidimensional, while artificial intelligence systems are usually designed over a set of real numbers, therefore trying to represent high-dimensional samples with mono-dimensional parameters. Therefore, data dimensions with a strong physical nature and complex relations are processed as sets of decorrelated components, losing crucial information and breaking the original structure of the data itself. As a consequence, several approaches in the literature try to overcome a bad data representation by enlarging model size up to billions of parameters, leading to networks with high-demanding computational resources and overfitting problems. An elegant and powerful method for solving these limitations is involving higher-order algebras, such as the quaternion one, when defining model operations and data representation. Therefore, in this work, we first focus on proposing novel and efficient quaternion architectures for modeling 3D audio data, with innovative representations that exploit a 6DoF characterization thanks to a dual quaternion formulation, which we also prove to be translation invariant. However, although quaternion neural networks are powerful methods for specific data types such as 3D audio or 3D human poses, they are limited to 4D inputs due to the four-dimensional nature of the quaternion number. Hence, we focus on generalizing this approach to a wider set of data, helping the spread of hypercomplex neural networks to various fields of application. Indeed, we present a framework to generalize quaternion neural networks to single-channel images, building multi-frequency samples leveraging the potentiality of quaternion wavelet transform to extract low- and high-frequency salient features from rich-of-detail images. This allows building a four-dimensional sample to be fed to quaternion neural networks, improving their performance. More interestingly, the enhanced sample can be also involved in conventional real-valued models, outperforming standard methods in a variety of tasks and datasets. In this study, we also introduce a novel method to go beyond quaternion and Clifford algebras. The proposed family of parameterized hypercomplex neural networks (PHNNs) is based on an innovative hypercomplex layer that can be defined in any desired domain, regardless of whether the algebra rules are preset. This allows the processing of any $n$D input, generalizing and preserving the advantages of sharing and reducing parameters. Our method grasps the convolution rules and the filter organization directly from data without requiring a rigidly predefined domain structure to follow. Such a malleability allows processing multidimensional inputs in their natural domain without annexing further dimensions, as done, instead, in quaternion neural networks for 3D inputs like color images. As a result, the proposed family of PHNNs operates with $1/n$ free parameters as regards its analog in the real domain. We demonstrate the versatility of this approach to multiple domains of application by performing experiments on various image, audio, and medical datasets in which our method outperforms real-, complex-, and quaternion-valued counterparts. Finally, we focus on involving the proposed approaches in the field of generative modeling, whose models are impressively increasing in size without carefully designing high-dimensional data representations. Indeed, latent relations among input dimensions, the RGB channels of an image, are essential to generate proper and good-quality samples. Simultaneously, in the last years, state-of-the-art generative models have received increasing interest and have known a boost in performance due also to the incredible growth of model size. As a side effect, this kind of model has become almost unreproducible and not accessible to research groups with lower budgets. In this scenario, hypercomplex algebras may help generative modeling in learning a suitable data representation that can preserve model performance even when the number of parameters is crucially reduced. In order to validate our theoretical claims, we conduct an extensive experimental evaluation at different scales in several domains of application, such as image, audio, 3D human pose, and medical ones. We perform tests in a variety of tasks ranging from classification and segmentation, up to generation and image modality translation to strengthen our claims and to provide theoretical and practical guidelines that may be useful for the research community and for future developments of hypercomplex and generative models.

Produzione scientifica

11573/1709447 - 2024 - Attention-map augmentation for hypercomplex breast cancer classification
Lopez, Eleonora; Betello, Filippo; Carmignani, Federico; Grassucci, Eleonora; Comminiello, Danilo - 01a Articolo in rivista
rivista: PATTERN RECOGNITION LETTERS (Elsevier BV:PO Box 211, 1000 AE Amsterdam Netherlands:011 31 20 4853757, 011 31 20 4853642, 011 31 20 4853641, EMAIL: nlinfo-f@elsevier.nl, INTERNET: http://www.elsevier.nl, Fax: 011 31 20 4853598) pp. - - issn: 0167-8655 - wos: WOS:001241411900001 (0) - scopus: 2-s2.0-85191528613 (0)

11573/1695674 - 2023 - Semantic Communications Based on Adaptive Generative Models and Information Bottleneck
Barbarossa, Sergio; Comminiello, Danilo; Grassucci, Eleonora; Pezone, Francesco; Sardellitti, Stefania; Di Lorenzo, Paolo - 01a Articolo in rivista
rivista: IEEE COMMUNICATIONS MAGAZINE (IEEE / Institute of Electrical and Electronics Engineers Incorporated:445 Hoes Lane:Piscataway, NJ 08854:(800)701-4333, (732)981-0060, EMAIL: subscription-service@ieee.org, INTERNET: http://www.ieee.org, Fax: (732)981-9667) pp. 36-41 - issn: 0163-6804 - wos: (0) - scopus: 2-s2.0-85172784611 (4)

11573/1669169 - 2023 - Dual quaternion ambisonics array for six-degree-of-freedom acoustic representation
Grassucci, E.; Mancini, G.; Brignone, C.; Uncini, A.; Comminiello, D. - 01a Articolo in rivista
rivista: PATTERN RECOGNITION LETTERS (Elsevier BV:PO Box 211, 1000 AE Amsterdam Netherlands:011 31 20 4853757, 011 31 20 4853642, 011 31 20 4853641, EMAIL: nlinfo-f@elsevier.nl, INTERNET: http://www.elsevier.nl, Fax: 011 31 20 4853598) pp. 24-30 - issn: 0167-8655 - wos: WOS:000920683000001 (3) - scopus: 2-s2.0-85145774706 (4)

11573/1693469 - 2023 - GROUSE. A task and model agnostic wavelet-driven framework for medical imaging
Grassucci, Eleonora; Sigillo, Luigi; Uncini, Aurelio; Comminiello, Danilo - 01a Articolo in rivista
rivista: IEEE SIGNAL PROCESSING LETTERS (IEEE / Institute of Electrical and Electronics Engineers Incorporated:445 Hoes Lane:Piscataway, NJ 08854:(800)701-4333, (732)981-0060, EMAIL: subscription-service@ieee.org, INTERNET: http://www.ieee.org, Fax: (732)981-9667) pp. 1397-1401 - issn: 1070-9908 - wos: WOS:001086210700001 (0) - scopus: 2-s2.0-85174843341 (0)

11573/1693887 - 2023 - Hypercomplex multimodal emotion recognition from EEG and peripheral physiological signals
Lopez, E.; Chiarantano, E.; Grassucci, E.; Comminiello, D. - 04b Atto di convegno in volume
congresso: 2023 IEEE International Conference on Acoustics, Speech and Signal Processing Workshops, ICASSPW 2023 (Rhodes; Greece)
libro: ICASSPW 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing Workshops, Proceedings - (979-8-3503-0261-5)

11573/1693477 - 2023 - PHYDI. Initializing parameterized hypercomplex neural networks as identity functions
Mancanelli, M.; Grassucci, E.; Uncini, A.; Comminiello, D. - 04b Atto di convegno in volume
congresso: 33rd IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2023 (Rome; Italy)
libro: IEEE International Workshop on Machine Learning for Signal Processing, MLSP - (979-8-3503-2411-2)

11573/1679634 - 2023 - Exploring the vaccine conversation on TikTok in Italy: beyond classic vaccine stances
Parisi, Lorenza; Mulargia, Simone; Comunello, Francesca; Bernardini, Vittoria; Bussoletti, Arianna; Nisi, Carla Rita; Russo, Luisa; Campagna, Ilaria; Lanfranchi, Barbara; Croci, Ileana; Grassucci, Eleonora; Gesualdo, Francesco - 01a Articolo in rivista
rivista: BMC PUBLIC HEALTH (London: BioMed Central, 2001-) pp. 1-13 - issn: 1471-2458 - wos: WOS:000986365200004 (1) - scopus: 2-s2.0-85159739356 (1)

11573/1693480 - 2023 - StawGAN: Structural-Aware Generative Adversarial Networks for Infrared Image Translation
Sigillo, L.; Grassucci, E.; Comminiello, D. - 04b Atto di convegno in volume
congresso: 56th IEEE International Symposium on Circuits and Systems, ISCAS 2023 (Monterey, USA)
libro: Proceedings - IEEE International Symposium on Circuits and Systems - (978-1-6654-5109-3)

11573/1693467 - 2023 - Sailing the SeaFormer. A transformer-based model for vessel route forecasting
Sigillo, L.; Marzilli, A.; Moretti, D.; Grassucci, E.; Greco, C.; Comminiello, D. - 04b Atto di convegno in volume
congresso: 33rd IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2023 (Rome; Italy)
libro: IEEE International Workshop on Machine Learning for Signal Processing, MLSP - (979-8-3503-2411-2)

11573/1693475 - 2023 - Dual quaternion rotational and translational equivariance in 3D rigid motion modelling
Vieira, G.; Grassucci, E.; Valle, M. E.; Comminiello, D. - 04b Atto di convegno in volume
congresso: 33rd IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2023 (Rome; Italy)
libro: IEEE International Workshop on Machine Learning for Signal Processing, MLSP - (979-8-3503-2411-2)

11573/1634356 - 2022 - Efficient Sound Event Localization and Detection in the Quaternion Domain
Brignone, Christian; Mancini, Gioia; Grassucci, Eleonora; Uncini, Aurelio; Comminiello, Danilo - 01a Articolo in rivista
rivista: IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS. II, EXPRESS BRIEFS (Piscataway, NJ : Institute of Electrical and Electronics Engineers, c2004-) pp. 2453-2457 - issn: 1549-7747 - wos: WOS:000790814000017 (4) - scopus: 2-s2.0-85126716404 (8)

11573/1610924 - 2022 - Quaternion generative adversarial networks
Grassucci, Eleonora; Cicero, Edoardo; Comminiello, Danilo - 02a Capitolo o Articolo
libro: Generative Adversarial Learning: Architectures and Applications - (978-3-030-91389-2; 978-3-030-91390-8)

11573/1669173 - 2022 - Hypercomplex image- to- image translation
Grassucci, Eleonora; Sigillo, Luigi; Uncini, Aurelio; Comminiello, Danilo - 04b Atto di convegno in volume
congresso: 2022 International Joint Conference on Neural Networks, IJCNN 2022 (Padua; Italy)
libro: Proceedings of the International Joint Conference on Neural Networks - (978-1-7281-8671-9)

11573/1665641 - 2022 - PHNNs: Lightweight Neural Networks via Parameterized Hypercomplex Convolutions
Grassucci, Eleonora; Zhang, Aston; Comminiello, Danilo - 01a Articolo in rivista
rivista: IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (Piscataway, NJ : Institute of Electrical and Electronics Engineeers) pp. 1-13 - issn: 2162-237X - wos: WOS:000899972600001 (7) - scopus: 2-s2.0-85144770762 (12)

11573/1606209 - 2021 - A quaternion-valued variational autoencoder
Grassucci, E.; Comminiello, D.; Uncini, A. - 04b Atto di convegno in volume
congresso: 2021 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2021 (Toronto; Canada)
libro: ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings - (978-1-7281-7605-5)

11573/1606334 - 2021 - An information-theoretic perspective on proper quaternion variational autoencoders
Grassucci, E.; Comminiello, D.; Uncini, A. - 01a Articolo in rivista
rivista: ENTROPY (Basel : MDPI, 1999-) pp. 1-17 - issn: 1099-4300 - wos: WOS:000676553300001 (7) - scopus: 2-s2.0-85110766164 (11)

11573/1547542 - 2021 - Flexible generative adversarial networks with non-parametric activation functions
Grassucci, E.; Scardapane, S.; Comminiello, D.; Uncini, A. - 02a Capitolo o Articolo
libro: Smart Innovation, Systems and Technologies - (978-981-15-5092-8; 978-981-15-5093-5)

¬© Università degli Studi di Roma "La Sapienza" - Piazzale Aldo Moro 5, 00185 Roma