Estimación de la incertidumbre aleatoria y epistémica en modelos GAN para simulaciones de detectores de partículas

The project proposes to enhance the analysis of data from a Ring Imaging Cherenkov (RICH) particle detector at CERN and other models for other particle detectors as well by focusing on estimating uncertainty. This project is an extension of the currently ongoing project where we aim to implement uncertainty estimation methods for a Cramer GAN that simulates a RICH particle detector. The primary goal of this current project is to disentangle two distinct types of uncertainty: aleatoric and epistemic. Aleatoric uncertainty stems from inherent randomness in the data, while epistemic uncertainty arises from incomplete knowledge or model limitations. By accurately quantifying these uncertainties, the project aims to improve the reliability and interpretability of the detector's measurements.
The methodology involves employing advanced statistical techniques and machine learning models tailored to handle complex high-energy physics data. The project will leverage both Bayesian and non-Bayesian methods to address both epistemic and aleatoric uncertainty, allowing for a more nuanced understanding of the limitations and assumptions within the model.
The outcomes of this project are expected to benefit both fundamental research in machine learning and applied research aiming to improve the operational effectiveness of the models for the particle detectors. A more precise estimation of uncertainties will enable researchers to make informed decisions about the reliability of their findings, ultimately contributing to the advancement of our understanding of particle behavior.

Mejorar la seguridad del uso de los modelos de aprendizaje profundo utilizados para simular la salida de los subdetectores del LHCb estimando y desenredando con precisión las incertidumbres aleatorias y epistémicas, mejorando así la confiabilidad y la interpretabilidad de las mediciones.

Aplicaciones de la computación en distintos dominios científicos, tecnológicos, organizacionales y sociales. Teoría y metodologías en computación.