Choose timezone
Your profile timezone:
Powered by Indico
v3.3.5
The rapid advancement of artificial intelligence and machine learning (AI/ML) in High Energy Physics has opened exciting avenues in the search for Physics beyond the Standard Model (BSM). This talk will explore developments in AI/ML applications and techniques in three areas to look, and eventually find, BSM Physics. First, we highlight model-independent searches for new physics, emphasising anomaly detection methods that leverage semi-supervised learning to identify deviations from Standard Model predictions without relying on specific BSM assumptions. Second, we discuss AI/ML-driven black-box search approaches for exploring highly constrained BSM parameter spaces, demonstrating their ability to drastically improve sampling efficiency and, most importantly, reveal novel phenomenological scenarios that have been hitherto overlooked. Finally, we showcase the use of ML in the hunt for exotic dark matter candidates, focusing on microlensing events associated with extended dark objects, such as boson stars, in current and near future surveys. These examples illustrate the immense potential of AI/ML to not only accelerate but also enhance the scope of phenomenological studies and experimental searches for BSM Physics, and to help use getting closer to find new physics.
The ATLAS collaboration at CERN has pioneered several machine learning (ML) techniques to improve particle reconstruction and enhance the precision of data analysis in high-energy physics. By leveraging advanced ML methods, ATLAS researchers can efficiently identify and characterise complex particle events, leading to significant background rejection and producing precise results to test the Standard Model. This presentation highlights key advancements, challenges, and prospects of ML in the ATLAS experiment, showcasing its transformative impact on data analysis on the Higgs boson properties.
In recent years, machine learning (ML) and artificial intelligence (AI) have emerged as transformative tools across various scientific domains, notably redefining the methodologies employed in experimental particle physics. In this talk we will explores the integral role that ML and AI play in advancing experimental techniques and enhancing data analysis within the field of particle physics, where enormous but also complex datasets are the norm.
We start with an overview of the key ML and AI methodologies currently employed, such as supervised and unsupervised learning, neural networks, and deep learning frameworks. Special attention will be given to how these methods are tailored to meet the specific demands of particle physics experiments, including particle tracking, event classification, data reduction, simulation, and real-time data processing. We will review how AI algorithms assist in real-time data processing and decision-making during experiments, significantly improving the efficiency of data collection and the accuracy of particle detection and event reconstruction.
We will discuss the challenges faced with the implementation of ML/AI solutions in this highly specialised field, including issues related to data quality, computational resources, and the need for domain-specific adaptations. Finally, we will consider the future potential of these technologies, discussing upcoming innovations and the integration of more advanced AI systems that could further revolutionise experimental approaches in particle physics.