Davide Talon | Homepage

I am a PostDoc at Fondazione Bruno Kessler (FBK) working on efficient multi-modal learning.
I am part of Deep Visual Learning(DVL) unit, advised by Prof. Elisa Ricci and Dr. Yiming Wang.

Before joining FBK I obtained the PhD in Computer Vision at University of Genova and Italian Institute of Technology (IIT) under the supervision of Dr. Alessio Del Bue and Prof. Stuart James. During the PhD, I visited the University of Amsterdam to work with Prof. Sara Magliacane. I completed both the BS in Information Engineering and the MS in Computer Engineering at University of Padova.
CV / Twitter

Research interests

Machine Learning: deep learning, representation learning, vision-language models, causality.

News

[Nov 2024] Good news! Ours recent work “GANzzle++: Generative approaches for jigsaw puzzle solving as local to global assignment in latent spatial representations” has been accepted for publication on Pattern Recognition Letters!
[Oct 2024] Our work “One VLM to Keep it Learning: Generation and Balancing for Data-free Continual Visual Question Answering” has been accepted at WACV2025!
[Apr 2024] Happy to share that we are organizing a workshop at ECCV2024 on Green Foundation Models (GreenFOMO), see you in Milan!
[Apr 2024] Check out our recent work “Uncertainty-guided Open-Set Source-Free Unsupervised Domain Adaptation with Target-private Class Segregation”
[Apr 2024] I’ve joined the Deep Visual Learning Group at Fondazione Bruno Kessler (FBK). I am excited to be working with Elisa Ricci and Yiming Wang!
[March 2024] I’ve successfully defended my PhD thesis on “Structured Representation Learning for Visual Data”
[Jan 2024] Our recent work “Towards the Reusability and Compositionality of Causal Representations” has been accepted at CLeaR as an Oral!
[Oct 2023] Our recent work “Towards the Reusability and Compositionality of Causal Representations” has been accepted at CRL@NeurIPS23.
[Jan 2023] Visiting INDELab @UvA to work on Causal Representation Learning with Prof. Sara Magliacane

Publications

GANzzle++: Generative approaches for jigsaw puzzle solving as local to global assignment in latent spatial representations, D. Talon, A. Del Bue, S. James, Pattern Recognition Letters, 2024.
One VLM to Keep it Learning: Generation and Balancing for Data-free Continual Visual Question Answering, D. Das, D. Talon, Y. Wang, M. Mancini, E. Ricci, IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2025.
Uncertainty-guided Open-Set Source-Free Unsupervised Domain Adaptation with Target-private Class Segregation, M. Litrico, D. Talon, S. Battiato, A. Del Bue, M. V. Giuffrida, P. Morerio, arXiv preprint arXiv:2404.10574, 2024. [PDF]
Towards the Reusability and Compositionality of Causal Representations,
D. Talon, P. Lippe, S. James, A. Del Bue, S. Magliacane, Conference on Causal Learning and Reasoning, 2024. [PDF]
Key Design Choices in Source-Free Unsupervised Domain Adaptation: An In-depth Empirical Analysis, A. Maracani, R. Camoriano, E. Maiettini, D. Talon, L. Rosasco, L. Natale, arXiv preprint arXiv:2402.16090, 2024. [PDF]
Towards the Reusability and Compositionality of Causal Representations,
D. Talon, P. Lippe, S. James, A. Del Bue, S. Magliacane, NeurIPS Workshop on Causal Representation Learning, 2023. [PDF][Poster]
Key Design Choices for Double-Transfer in Source-Free Unsupervised Domain Adaptation, A. Maracani, R. Camoriano, E. Maiettini, D. Talon, L. Rosasco, L. Natale, ECML Workshop on Reliable Multimodal Learning Across Domains, 2023. [PDF]
GANzzle: Reframing jigsaw puzzle solving as a retrieval task using a generative mental image, D. Talon, A. Del Bue, S. James, IEEE International Conference on Image Processing (ICIP), 2022. [PDF] [Data] [Poster]
Comparing DASH Adaptation Algorithms in a Real Network Environment,
D. Talon, L. Attanasio, F. Chiariotti, M. Gadaleta, A. Zanella and M. Rossi, European Wireless 2019, Denmark. [PDF] [Code]

Service to the community

Workshop Organizer: GreenFOMO@ECCV24.
Conference Reviewer: BMVC21, ECCV24, IROS24.
Journal Reviewer: Pattern Recognition.
Workshop Reviewer: MULA@CVPR22, CRL@UAI22, VISART@ECCV22, nCSI@NeurIPS22, MULA@CVPR23, AtC@ECMLPKDD23, CRL@NeurIPS23, MULA@CVPR24.