By Marcos7 min
Who Will Dictate the Rules of the Game?
EN
The struggle for dominance in artificial intelligence.
#geopolitics
AI Researcher
Marcos López de Castro
Physicist turned trustworthy AI researcher. Specialized in conformal prediction and uncertainty quantification, I am particularly interested in knowledge discovery, digital medicine, mathematical modelling, and the potential of classical and quantum computing.
External links
Featured posts
Generated by AI, reviewed by Marcos9 min
Calibration and Uncertainty in Machine Learning: Advances, Limits, and Current Challenges
EN
Calibration and uncertainty quantification in machine learning have made great strides, yet they remain far from offering universal guarantees outside of stable environments. This analysis reviews their current state, the most common methods, and their main limitations.
#machine learning#calibration#uncertainty#uncertainty quantification
Join the blog
As a researcher, I know it's impossible to read everything that's published. That's why I've developed an agent-based AI system that scans and screens on my behalf the latest material relevant to me. Subscribe to receive these findings periodically, along with the occasional personal reflection.
Latest News
Conference
Participation in AIME 2026
From July 7 to 10, 2026, I participated in the 24th International Conference on Artificial Intelligence in Medicine (AIME 2026), hosted by the University of Ottawa. We presented "Bidirectional Floating Feature Selection Guided by Uncertainty Quantification", a study introducing Conformal Bidirectional Floating Search (CBFS). The method uses conformal prediction to find compact feature subsets that reduce predictive uncertainty and was evaluated on synthetic data and multi-cohort transcriptomic data from immunotherapy studies. The work highlights the potential of uncertainty-aware feature selection for reliable biomarker discovery and is published in Springer's Lecture Notes in Artificial Intelligence proceedings.
Invited Talk
Invited lecture
On February 27, 2026, I had the pleasure of being an invited speaker in the University Master's in Mathematical Modeling at the University of Salamanca (USAL). During the two-hour session, I delivered the lecture titled "Does your model know what it doesn't know? Conformal prediction for reliable uncertainty quantification". It was a great experience to share ideas on the importance of quantifying uncertainty reliably in machine learning models.
Research Stay
Research stay completed at Ghent University
Completed a collaborative research stay on robust uncertainty estimation and benchmark transferability.
Press
Diario de Navarra covers the IFIT index for more personalized cancer therapies
Diario de Navarra featured the presentation of the IFIT index, an AI tool designed to assess the immunological fitness of oncology patients and support more personalized therapeutic decisions.