Gianvito Losapio

Gianvito Losapio

Artificial I.

About me

Hi! I am 27 and have been into AI research for 5+ years now. I am currently a first-year PhD student at Politecnico di Milano, working on reinforcement learning in Marcello Restelli‘s lab (consistently well-represented at NeurIPS, ICML). My PhD is funded by the AI4REALNET project.

Previously, I was at the Machine Learning Genoa Center in the group of Lorenzo Rosasco (also at MIT) and Agnese Seminara.

🧠 Intelligence 🧠 to me is just a computational problem. I am working on a formal theory out of this idea. Reach out if you are interested, I am looking for collaborators!

During formative years I interned at CERN and at the National Research Council of Italy. I also attended two editions of the M2L school (sponsored by Google DeepMind) presenting two posters.

Interests

  • Reinforcement learning
  • Language and evolution

Education

  • M.S. in Artificial Intelligence, 2022

  • B.S. in Computer Science and Automation Engineering, 2019

Publications

(2024). State and Action Factorization in Power Grids. ECML-PKDD 2024, Machine Learning for Sustainable Power Systems workshop.

PDF Poster Slides DOI

(2022). Learning new physics efficiently with nonparametric methods. European Physical Journal C, 82(10), 879.

PDF Poster DOI

(2022). Smart Balancing of E-scooter Sharing Systems via Deep Reinforcement Learning: a preliminary study. WOA2021 & Intelligenza Artificiale, 16(1), 49-67.

PDF DOI

(2022). Efficient Machine Learning for new physics discoveries. Master Thesis.

PDF Slides

(2021). Efficient kernel methods for model-independent new physics searches. NeurIPS 2021, Machine Learning and the Physical Sciences Workshop.

PDF

(2021). Machine Learning for 40 MHZ Scouting at CMS. CERN Technical Report, Zenodo.

PDF Slides Video DOI

(2020). Lightweight and efficient convolutional neural networks for recognition of dolphin dorsal fins. IMEKO TC-19 International Workshop on Metrology for the Sea.

PDF Video

(2020). Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins. Electronics, 9(5), 758.

PDF Poster DOI

(2019). Tecniche di Deep Learning e Object Detection per la foto-identificazione dei cetacei. Bachelor Thesis.

PDF Slides

PhD activities

Master Thesis co-supervision
Oct 10, 2024
Italian Tech Week
Sep 25, 2024 Turin (Italy)
Talk + poster @ ECML-PKDD 2024
Sep 9, 2024 Vilnius (Lithuania)
UCL NeuroAI annual event
Jul 15, 2024 London (UK)
Oxford Machine Learning summer school
Jul 10, 2024 Oxford (UK)
AI4REALNET annual meeting
Jun 12, 2024 Delft (the Netherlands)
AI4REALNET kick-off meeting
Oct 11, 2023 Porto (Portugal)

Misc

Conversation with Professor Alessandro Verri

During my period at MaLGa, I had the pleasure to speak with professor Alessandro Verri about topics such as artificial intelligence, faith, consciousness. Beyond his exceptional career, his humanity has impressed me ever since the first time I met him in his office (he invited me because I wanted to ask some questions about the new Master in AI that was starting in Genoa later that year, 2019).
Last updated on Oct 14, 2024 17 min read Interviews

Notes

Efficient generalization in Partially Observable Markov Decision Processes: a case study

Contact

  • gvlosapio (AT) gmail (DOT) com
  • Via Camillo Golgi 39, Milan | Building 21 - Office 15 (first floor),