Speaker
Description
In the realm of High-Energy Physics (HEP), the quest for new physics phenomena demands innovative approaches to data analysis. Anomaly detection (AD) tools, traditionally employed in fields such as cybersecurity and industrial quality control, are emerging as potent resources for uncovering rare and unexpected signals amidst vast datasets generated by particle colliders. By leveraging machine learning algorithms, statistical methods, and advanced data preprocessing techniques, anomaly detection tools offer a promising avenue for identifying elusive signals of new physics beyond the Standard Model. Notably, one distinct advantage of AD tools is their model independence, enabling the exploration of signals not constrained by any theoretical framework.
This study delves into the transformative potential of anomaly detection tools as innovative analytical resources within the domain of HEP. Through a detailed case study utilizing auto-encoders for a Run-2 physics analysis conducted with ATLAS data, we aim to showcase the efficacy and versatility of AD tools in uncovering novel physics phenomena. Additionally, we examine the potential benefits and challenges associated with these approaches.
