"Across vastly different scales of physics - from quantum to astronomical - a common analytical challenge emerges: extracting precise physical parameters from faint signals embedded in noisy timeseries data. In this talk, we explore how shared methodological innovations, particularly machine learning, are transforming our ability to tackle this challenge and enabling a new era of precision science.
Using the Laser Interferometer Gravitational-Wave Observatory (LIGO), which probes the universe through gravitational waves, and Project 8, which seeks to determine the neutrino mass through precision spectroscopy of tritium beta decay, as case studies, we draw connections not only across scales but also across disciplines, bridging particle physics and astrophysics through their shared reliance on timeseries analysis. The data encountered in these experiments pose fundamental challenges even to modern AI: long-range temporal dependencies, low signal-to-noise ratios, and high sampling rates push standard architectures to their limits. We present how recent advances in sequence modeling address these challenges and offer powerful, transferable solutions that can drive precision science across domains.
Rather than treating each experiment in isolation, we highlight the interconnection of AI-driven analysis tools between particle physics and astrophysics.
We find the biggest and the smallest scales in physics have more in common than one might expect - and the tools we develop may well unlock fundamental physical discoveries across all scales while bringing novel AI strategies to the forefront."
