Astronomaly Protege: Discovery through Human-machine Collaboration
Modern telescopes generate catalogs of millions of objects with the potential for new scientific discoveries, but this is beyond what can be examined visually. Here we introduce ASTRONOMALY: PROTEGE, an extension of the general-purpose machine-learning-based active anomaly detection framework ASTRON...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IOP Publishing
2025-01-01
|
| Series: | The Astronomical Journal |
| Subjects: | |
| Online Access: | https://doi.org/10.3847/1538-3881/ada14c |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832540891804336128 |
|---|---|
| author | Michelle Lochner Lawrence Rudnick |
| author_facet | Michelle Lochner Lawrence Rudnick |
| author_sort | Michelle Lochner |
| collection | DOAJ |
| description | Modern telescopes generate catalogs of millions of objects with the potential for new scientific discoveries, but this is beyond what can be examined visually. Here we introduce ASTRONOMALY: PROTEGE, an extension of the general-purpose machine-learning-based active anomaly detection framework ASTRONOMALY. PROTEGE is designed to provide well-selected recommendations for visual inspection, based on a small amount of optimized human labeling. The resulting sample contains rare or unusual sources that are simultaneously as diverse as the human trainer chooses and of scientific interest to them. We train PROTEGE on images from the MeerKAT Galaxy Cluster Legacy Survey, leveraging the self-supervised deep learning algorithm Bootstrap Your Own Latent to find a low-dimensional representation of the radio galaxy cutouts. By operating in this feature space, PROTEGE is able to recommend interesting sources with completely different morphologies in image space to those it has been trained on. This provides important advantages over similarity searches, which can only find more examples of known sources, or blind anomaly detection, which selects unusual but not necessarily scientifically interesting sources. Using an evaluation subset, we show that, with minimal training, PROTEGE provides excellent recommendations and find that it is even able to recommend sources that the authors missed. We briefly highlight some of PROTEGE's top recommendations, which include X- and circular-shaped sources, filamentary structures, and one-sided structures. These results illustrate the power of an optimized human-machine collaboration, such as PROTEGE, to make unexpected discoveries in samples beyond human-accessible scales. |
| format | Article |
| id | doaj-art-4ff8417c4d0349fda0dde7117fb55de4 |
| institution | Kabale University |
| issn | 1538-3881 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| series | The Astronomical Journal |
| spelling | doaj-art-4ff8417c4d0349fda0dde7117fb55de42025-02-04T13:20:16ZengIOP PublishingThe Astronomical Journal1538-38812025-01-01169312110.3847/1538-3881/ada14cAstronomaly Protege: Discovery through Human-machine CollaborationMichelle Lochner0https://orcid.org/0000-0003-2221-8281Lawrence Rudnick1https://orcid.org/0000-0001-5636-7213Department of Physics and Astronomy, University of the Western Cape , Bellville, Cape Town, 7535, South Africa ; dr.michelle.lochner@gmail.com; South African Radio Astronomy Observatory , Liesbeek House, River Park, Liesbeek Parkway, Mowbray 7705, South AfricaMinnesota Institute for Astrophysics, University of Minnesota , 116 Church Street SE, Minneapolis, MN 55455, USAModern telescopes generate catalogs of millions of objects with the potential for new scientific discoveries, but this is beyond what can be examined visually. Here we introduce ASTRONOMALY: PROTEGE, an extension of the general-purpose machine-learning-based active anomaly detection framework ASTRONOMALY. PROTEGE is designed to provide well-selected recommendations for visual inspection, based on a small amount of optimized human labeling. The resulting sample contains rare or unusual sources that are simultaneously as diverse as the human trainer chooses and of scientific interest to them. We train PROTEGE on images from the MeerKAT Galaxy Cluster Legacy Survey, leveraging the self-supervised deep learning algorithm Bootstrap Your Own Latent to find a low-dimensional representation of the radio galaxy cutouts. By operating in this feature space, PROTEGE is able to recommend interesting sources with completely different morphologies in image space to those it has been trained on. This provides important advantages over similarity searches, which can only find more examples of known sources, or blind anomaly detection, which selects unusual but not necessarily scientifically interesting sources. Using an evaluation subset, we show that, with minimal training, PROTEGE provides excellent recommendations and find that it is even able to recommend sources that the authors missed. We briefly highlight some of PROTEGE's top recommendations, which include X- and circular-shaped sources, filamentary structures, and one-sided structures. These results illustrate the power of an optimized human-machine collaboration, such as PROTEGE, to make unexpected discoveries in samples beyond human-accessible scales.https://doi.org/10.3847/1538-3881/ada14cAstronomy data analysisAstronomy softwareOpen source softwareExtragalactic radio sourcesRadio active galactic nuclei |
| spellingShingle | Michelle Lochner Lawrence Rudnick Astronomaly Protege: Discovery through Human-machine Collaboration The Astronomical Journal Astronomy data analysis Astronomy software Open source software Extragalactic radio sources Radio active galactic nuclei |
| title | Astronomaly Protege: Discovery through Human-machine Collaboration |
| title_full | Astronomaly Protege: Discovery through Human-machine Collaboration |
| title_fullStr | Astronomaly Protege: Discovery through Human-machine Collaboration |
| title_full_unstemmed | Astronomaly Protege: Discovery through Human-machine Collaboration |
| title_short | Astronomaly Protege: Discovery through Human-machine Collaboration |
| title_sort | astronomaly protege discovery through human machine collaboration |
| topic | Astronomy data analysis Astronomy software Open source software Extragalactic radio sources Radio active galactic nuclei |
| url | https://doi.org/10.3847/1538-3881/ada14c |
| work_keys_str_mv | AT michellelochner astronomalyprotegediscoverythroughhumanmachinecollaboration AT lawrencerudnick astronomalyprotegediscoverythroughhumanmachinecollaboration |