Giovani Valdrighi
Campinas, Brazil
I’m a third-year Ph.D. candidate in Computer Science at the Institute of Computing at the University of Campinas (IC - UNICAMP). I’m a member of the Hub of Artificial Intelligence and Cognitive Architectures (HIAAC), performing research in Trusthyworth AI, mainly in the area of fairness and explainability. I hold a MSc in Mathematical Modelling at the School of Applied at FGV and BSc degree in Applied Math from the same institution. In the last few years, I have been a researcher at the Visual Data Science Lab(VDS), working on visualization techniques for spatiotemporal data. I also had the opportunity to be a visiting researcher in the VIDA Lab at NYU Tandon, working with deep learning techniques to predict socioeconomic indexes from remote sensing data.
news
| May 24, 2026 | My paper “Long-term Fairness with Selective Labels” was accepted for publication at ICML 2026. I’ll be at South-Korea to present this work! |
|---|---|
| Apr 26, 2026 | I’ll be presented two posters titled “Long-term Fairness with Selective Labels” and “Navigating the Rashomon Set: The Impact of Score Distributions and Decision Thresholds on Model Agreement” at the Algorithmic Fairness Across Alignment Procedures and Agentic Systems workshop at ICLR 2026. |
| Mar 23, 2026 | I participed in the organization of the H.IAAC workshop that took place at UNICAMP on March 23rd-25th, 2026. More than 150 participants attended the event. I also presented our work at the NeurIPS 2025 competition, “Early Training Evaluation of Language Models”. |
| Dec 07, 2025 | My colleagues and I won the first-place prize at the NeurIPS 2025 competition, “Early Training Evaluation of Language Models” (E2LM). |
| Nov 18, 2025 | I participed in the organization of the Workshop for Thesis and Dissertations (WTD) that took place at the Institute of Computing at UNICAMP on November 18th-19th, 2025. I also presented my work “Long-term Fairness with Selective Labels”. |
selected publications
- MoReVis: A Visual Summary for Spatiotemporal Moving RegionsIEEE Transactions on Visualization and Computer Graphics, 2023
- M^2FGB: A Min-Max Gradient Boosting Framework for Subgroup Fairness2025