Uncertainty quantification (UQ) is a vast field of research which has received the attention of mathematicians and physicists alike. UQ goes beyond acknowledging the limitations of models and data: it enables the constructions of trustworthy and reliable predictive models, key components of many real-world applications.
UQ4ML is the meeting point of different research directions, from applied mathematics and statistics to theoretical and experimental physics. We shall discuss recent advancements in data analysis, with a strong focus on particle physics. Discussion will span across the domain of machine learning and AI, as well as fundamentals in statistics.
By bringing together experts from these diverse fields, we aim to foster discussion, innovative solutions, and collaborations that will drive the future of UQ and its applications.
-
Simulations and Coding
This session presents an overview of uncertainty quantification techniques for simulations in machine learning. By examining how uncertainties propagate through complex models, the session explores the reliability and limitations of their results. This knowledge is crucial for advancing the field and ensuring that machine learning applications are robust and trustworthy.
-
HEP - Theory
This session explores the theoretical aspects of high energy physics through the lens of machine learning. By addressing the unique challenges and uncertainties in this domain, we review innovative approaches to enhance our understanding of fundamental physics. This intersection of fields highlights the broader implications of uncertainty quantification in scientific discovery.
-
HEP - Experiment
This session delves into the practical challenges and issues of applying AI in experimental high energy physics. Seminars will deal with how uncertainty quantification plays a pivotal role in ensuring the accuracy and reliability of experimental results. This discussion is essential for advancing both the field of high energy physics and the broader application of machine learning in experimental sciences.
-
Data Analysis, Time series, Causal analysis
This session focuses on the critical aspects of data analysis in time series, though it is not limited to it. Understanding and quantifying uncertainty in time-dependent data is essential for accurate forecasting and decision-making. Additionally, exploring causal relationships helps in identifying true drivers of outcomes, rather than mere correlations, which is vital for robust machine learning models.
-
Deep Learning and Uncertainty Quantification
Deep learning models have shown remarkable success in various domains, but they often lack transparency in their predictions. This session addresses the integration of uncertainty quantification techniques with deep learning, enabling more reliable and interpretable models.