From Uncertainty to Discovery: Uncertainty Quantification at the Frontier of BSM Physics

Mapping the ever-growing theoretical ecosystem of beyond standard model (BSM) theories and classifying various parameter configurations in a dimensionally reduced, human-readable space, is crucial for identifying high-impact measurements in future collider searches for new physics. In this setting, we leverage recent machine learning developments in evidential deep learning (EDL) for uncertainty quantification (UQ) to define quantitative metrics of assessing overlaps and discrepancies between various models. Combined with the power of generative AI, we can reconstruct plausible new theories representing these overlapped regions. I will present an ongoing development of these methodologies aimed towards enhancing the fidelity of phenomenological investigations of BSM physics. This approach outlines a strategic framework to inform experimental designs, increasing the discovery potential at the frontiers of particle physics.