TIDMAD: Time Series Dataset for Discovering Dark Matter with AI Denoising

TIDMAD is the first dataset and benchmark from a dark matter physics experiment, providing ultra-long time series data and comprehensive tools that enable machine learning models to directly advance the fundamental physics search for dark matter.

This data is availble for download via download_data.py. Metadata for this dataset is specified in TIDMAD_croissant.json. The file names are listed in filelist.dat. For furhter information and publically available code, please see the associated GitHub repository. For more information on this dataset and benchmark, please reference our TIDMAD paper.