Valentino Maiorca

My biography?

It's right there!

Towards a more general understanding of the latent shape of information I research how similar latent representations emerge in different neural networks, both artificial and biological, and how we can leverage this understanding for various applications.

I focus on intrinsic factors from input data (its semantics) revealing how networks process similar concepts, and extrinsic factors, like random initialization or data modality, which alter the space where these representations exist rathern than their structure.

My work involves formalizing “latent communication” to align different neural networks’ latent spaces. This has practical applications in deep learning model stitching, single-cell alignment in biology, and brain decoding in neuroscience.

By understanding the universal principles of data processing and the geometry of latent spaces, my work aims to drive advancements across multiple fields and positively impact both academic research and real-world applications.

Full CV available here.

Selected Publications

COSYNE

Multi-subject neural decoding via relative representations

Valentino Maiorca, Simone Azeglio, Marco Fumero, Clémentine Dominé, Emanuele Rodolà, and Francesco Locatello

In Cosyne Abstracts, 2024

ICLR

Relative representations enable zero-shot latent space communication

Luca Moschella, Valentino Maiorca, Marco Fumero, Antonio Norelli, Francesco Locatello, and Emanuele Rodolà

In The Eleventh International Conference on Learning Representations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023, 2023

arXiv Bib Code

@inproceedings{relrep,
  author = {Moschella, Luca and Maiorca, Valentino and Fumero, Marco and Norelli, Antonio and Locatello, Francesco and Rodol{\`{a}}, Emanuele},
  title = {Relative representations enable zero-shot latent space communication},
  booktitle = {The Eleventh International Conference on Learning Representations,
                   {ICLR} 2023, Kigali, Rwanda, May 1-5, 2023},
  publisher = {OpenReview.net},
  year = {2023},
  url = {https://openreview.net/pdf?id=SrC-nwieGJ},
  timestamp = {Fri, 30 Jun 2023 14:55:52 +0200},
  biburl = {https://dblp.org/rec/conf/iclr/MoschellaMFNLR23.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org},
}

NeurIPS

ASIF: Coupled Data Turns Unimodal Models to Multimodal without Training

Antonio Norelli, Marco Fumero, Valentino Maiorca, Luca Moschella, Emanuele Rodolà, and Francesco Locatello

In Thirty-seventh Conference on Neural Information Processing Systems, 2023

arXiv Bib Code

@inproceedings{norelli2023asif,
  title = {{ASIF}: Coupled Data Turns Unimodal Models to Multimodal without Training},
  author = {Norelli, Antonio and Fumero, Marco and Maiorca, Valentino and Moschella, Luca and Rodol{\`a}, Emanuele and Locatello, Francesco},
  booktitle = {Thirty-seventh Conference on Neural Information Processing Systems},
  year = {2023},
  url = {https://openreview.net/forum?id=XjOj3ZmWEl},
}

ACL
Accelerating Transformer Inference for Translation via Parallel Decoding

Andrea Santilli, Silvio Severino, Emilian Postolache, Valentino Maiorca, Michele Mancusi, Riccardo Marin, and Emanuele Rodolà

In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Jul 2023

Abs Bib

Autoregressive decoding limits the efficiency of transformers for Machine Translation (MT). The community proposed specific network architectures and learning-based methods to solve this issue, which are expensive and require changes to the MT model, trading inference speed at the cost of the translation quality. In this paper, we propose to address the problem from the point of view of decoding algorithms, as a less explored but rather compelling direction. We propose to reframe the standard greedy autoregressive decoding of MT with a parallel formulation leveraging Jacobi and Gauss-Seidel fixed-point iteration methods for fast inference. This formulation allows to speed up existing models without training or modifications while retaining translation quality. We present three parallel decoding algorithms and test them on different languages and models showing how the parallelization introduces a speedup up to 38% w.r.t. the standard autoregressive decoding and nearly 2x when scaling the method on parallel resources. Finally, we introduce a decoding dependency graph visualizer (DDGviz) that let us see how the model has learned the conditional dependence between tokens and inspect the decoding procedure.
@inproceedings{santilli-etal-2023-accelerating, title = {Accelerating Transformer Inference for Translation via Parallel Decoding}, author = {Santilli, Andrea and Severino, Silvio and Postolache, Emilian and Maiorca, Valentino and Mancusi, Michele and Marin, Riccardo and Rodol{\`{a}}, Emanuele}, editor = {Rogers, Anna and Boyd-Graber, Jordan and Okazaki, Naoaki}, booktitle = {Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)}, month = jul, year = {2023}, address = {Toronto, Canada}, publisher = {Association for Computational Linguistics}, url = {https://aclanthology.org/2023.acl-long.689}, doi = {10.18653/v1/2023.acl-long.689}, pages = {12336--12355}, }

NeurIPS

Latent Space Translation via Semantic Alignment

Valentino Maiorca, Luca Moschella, Antonio Norelli, Marco Fumero, Francesco Locatello, and Emanuele Rodolà

In Thirty-seventh Conference on Neural Information Processing Systems, Jul 2023

arXiv Bib

@inproceedings{maiorca2023latent,
  title = {Latent Space Translation via Semantic Alignment},
  author = {Maiorca, Valentino and Moschella, Luca and Norelli, Antonio and Fumero, Marco and Locatello, Francesco and Rodol{\`a}, Emanuele},
  booktitle = {Thirty-seventh Conference on Neural Information Processing Systems},
  year = {2023},
}

ICLR

From Bricks to Bridges: Product of Invariances to Enhance Latent Space Communication

Irene Cannistraci, Luca Moschella, Marco Fumero, Valentino Maiorca, and Emanuele Rodolà

In The Twelfth International Conference on Learning Representations, Jul 2024

Bib

@inproceedings{cannistraci2024from,
  title = {From Bricks to Bridges: Product of Invariances to Enhance Latent Space Communication},
  author = {Cannistraci, Irene and Moschella, Luca and Fumero, Marco and Maiorca, Valentino and Rodol{\`a}, Emanuele},
  booktitle = {The Twelfth International Conference on Learning Representations},
  year = {2024},
  url = {https://openreview.net/forum?id=vngVydDWft},
}