Published May 19, 2025 | Version 1.0.0
Dataset Open

ExoMiner++: Enhanced Transit Classification and a New Vetting Catalog for 2-Minute TESS Data

  • 1. Universities Space Research Association (NASA ARC)
  • 2. ROR icon Ames Research Center

Description

Summary

This data repository is associated with the work presented in paper "ExoMiner++: Enhanced Transit Classification and a New Vetting Catalog for 2-Minute TESS Data" .

Contents

  • exominerplusplus_catalog_unk_tces_s1-s67_tess-spoc-2min_complete_1-16-2025_1014.csv: vetting catalog of unlabeled (UNK) TESS SPOC 2-min TCEs for sector runs S1-S67 (included multi-sector runs) produced using the ExoMiner++ models trained for the experiment "TESS+Kepler" (see Table 6 in paper). A shorter catalog is available online here (TCEs with score < 0.1 are excluded due to memory constraints).
  • exominerplusplus_catalog_labeled_tces_s1-s67_tess-spoc-2min_complete_1-14-2025_1039.csv: catalog for the labeled TESS SPOC 2-min TCEs for sector runs S1-S67 (included multi-sector runs) dataset used to train the models in the experiment "TESS+Kepler" (see Table 6 in paper).
  • exominer_plusplus_architecture.png: image of ExoMiner++ architecture.
  • tfrecords_tess-spoc-2min_s1-s67_9-24-2024_1159.tar.xz: a compressed file of the TESS dataset used to train and evaluate the models in the paper (does not include the Kepler data). The examples are split across multiple files (aka shards) in TFRecord format. This dataset was used as source for the creation of the cross-validation dataset used in the paper. The features are unnormalized (at least the ones that require normalization using training set statistics). You would first split the dataset into, for example, a single train-val-test split OR into multiple cross-validation folds; then compute normalization statistics; and then normalize the features accordingly. This dataset includes both unlabeled examples (aka UNK) and the remaining labeled ones (all other labels) - see feature attribute 'label' after parsing an example from the TFRecord dataset. Includes an auxiliary table named "shards_tbl.csv" that describes the set of examples in the TFRecord dataset, with information about which file they are stored in (column 'shard'), and their order in the shard file (column 'example_i_tfrec').

Related Software

These data are associated with version v1.0.0 (ExoMiner++) of ExoMiner found under NASA GitHub.

Files

exominer_plusplus_architecture.png

Additional details

Related works

Continues
Journal: 10.3847/1538-4357/ac4399 (DOI)
Is supplement to
Preprint: arXiv:2502.09790 (arXiv)
Journal: 10.3847/1538-3881/ae03a4 (DOI)

Funding

National Aeronautics and Space Administration
TESS XRP 2022704 22-XRP22 2-0173
National Aeronautics and Space Administration
Intelligent Systems Research and Development-707 3 (ISRDS-3) 80ARC020D0010

Software

Repository URL
https://github.com/nasa/ExoMiner/tree/main
Programming language
Python , HTML , CSS , Shell
Development Status
Active