Research & publications
Geometry, graphs
& generative models
My research lies at the intersection of computer vision, geometric deep learning and graph learning. I develop models that use geometric and relational structure, with applications spanning 3D data, music, healthcare and molecular design.
Recent publications 2023–2026
Journal, conference and workshop papers from 2023 onwards, ordered by year and then alphabetically by title. Conference papers are grouped by event year; journal articles by issue year. Preprint versions of published papers are not listed twice.
19 publications
2026
Diagnosing Alzheimer’s disease using hypergraph neural networks with prompt tuning
Pattern Recognition 2026
Hypergraph learning and prompt tuning for Alzheimer’s disease classification.
GraFix++: A novel graph transformer based on a fixed multi-head structural attention mechanism
Pattern Recognition 2026
A graph transformer built around fixed structural attention.
Leveraging Generative Models for Uncertainty-aware Sea Waves 3D Reconstruction
IEEE TGRS 2026
Generative modelling for 3D sea-surface reconstruction with uncertainty.
2025
COCOLA: Coherence-Oriented Contrastive Learning of Musical Audio Representations
ICASSP 2025
Learning representations of harmonic and rhythmic coherence to evaluate musical accompaniments.
Generating Graphs via Spectral Diffusion
ICLR 2025
Generating graph structures by learning to sample Laplacian eigenvalues and eigenvectors.
Graph Kernel Neural Networks
IEEE TNNLS 2025
Graph kernels as a building block for neural networks on irregular graph-structured data.
Naturalistic Music Decoding from EEG Data via Latent Diffusion Models
ICASSP 2025
Exploring latent diffusion models for music reconstruction from non-invasive EEG recordings.
STAGE: Stemmed Accompaniment Generation through Prefix-Based Conditioning
ISMIR 2025
Single-instrument accompaniment generation conditioned on an existing musical mixture.
2024
Generalized Multi-Source Inference for Text Conditioned Music Diffusion Models
ICASSP 2024
Compositional music generation and source separation with text-conditioned diffusion models.
GNN-LoFI: A novel graph neural network through localized feature-based histogram intersection
Pattern Recognition 2024
Learning from local distributions of node features using histogram-intersection kernels.
GraFix: A Graph Transformer with Fixed Attention Based on the WL Kernel
ICPR 2024 · proceedings 2025
Fixed structural attention in graph transformers using the Weisfeiler–Lehman kernel.
LESI-GNN: An Interpretable Graph Neural Network Based on Local Structures Embedding
S+SSPR 2024 · proceedings 2025
Interpretable graph learning through embeddings of local structures.
Multi-Source Diffusion Models for Simultaneous Music Generation and Separation
ICLR 2024
A single diffusion model for music generation, source separation and source imputation.
SelfGeo: Self-supervised and Geodesic-Consistent Estimation of Keypoints on Deformable Shapes
ECCV 2024
Self-supervised keypoint estimation that preserves geodesic consistency across deformable shapes.
2023
Differentiable Graph Module (DGM) for Graph Convolutional Networks
IEEE TPAMI 2023
Learning the graph connectivity alongside a downstream task, including when the underlying graph is unknown.
Graph-in-Graph (GiG): Learning interpretable latent graphs in non-Euclidean domain for biological and healthcare applications
Medical Image Analysis 2023
Learning latent relationships between graph-valued samples for biological and healthcare applications.
Guided diffusion for inverse molecular design
Nature Computational Science 2023
Guided diffusion for molecular design, combining graph-based property prediction with generative modelling.
Latent Autoregressive Source Separation
AAAI 2023
Source separation with pretrained autoregressive models in a quantized latent space.
Spectral Maps for Learning on Subgraphs
NeurReps workshop 2023 · proceedings 2024
Spectral representations of graph-to-subgraph maps for graph learning.
No publications match these filters. Try another topic or search term.
Preprints
Listed separately from the published papers above.
FolAI: Synchronized Foley Sound Generation with Semantic and Temporal Alignment
arXiv preprint · 2024, revised 2025
Foley sound generation with semantic and temporal control.
Earlier selected work
Selected work on 3D geometry and graph learning from 2019–2020. The full bibliography is available on DBLP.
Generating Adversarial Surfaces via Band-Limited Perturbations
Computer Graphics Forum / SGP 2020
Smooth adversarial perturbations for deformable 3D surfaces and their use in adversarial training.
Latent-Graph Learning for Disease Prediction
MICCAI 2020
Learning population-graph connectivity for disease prediction with graph convolutional networks.
LIMP: Learning Latent Shape Representations with Metric Preservation Priors
ECCV 2020
Geometry-preserving latent representations for deformable 3D shapes.
The Average Mixing Kernel Signature
ECCV 2020
A spectral descriptor for non-rigid 3D shapes based on continuous-time quantum walks.
Isospectralization, or How to Hear Shape, Style, and Correspondence
CVPR 2019
Deforming shapes to align their Laplacian spectra, with applications to reconstruction and correspondence.
Academic service
Associate Editor
Pattern Recognition
Area Chair
NeurIPS 2024, 2025 and 2026
Vice Chair · 2024–2026
IAPR TC2 — Structural and Syntactical Pattern Recognition
Industry collaboration
My industry experience includes scientific advice on computer-vision-based quality inspection and contactless measurement with the Ca’ Foscari spin-off DigitalViews.
Thesis archive
Master’s thesis (Italian, PDF) — Corrispondenza densa tra forme attraverso embedding di superficie e correlazione sferica.