Finding radio transients with anomaly detection and active learning based on volunteer classifications
Bibliographic record
Abstract
ABSTRACT In this work, we explore the applicability of unsupervised machine learning algorithms to finding radio transients. Facilities such as the Square Kilometre Array (SKA) will provide huge volumes of data in which to detect rare transients; the challenge for astronomers is how to find them. We demonstrate the effectiveness of anomaly detection algorithms using 1.3 GHz light curves from the SKA precursor MeerKAT. We make use of three sets of descriptive parameters (‘feature sets’) as applied to two anomaly detection techniques in the astronomaly package and analyse our performance by comparison with citizen science labels on the same data set. Using transients found by volunteers as our ground truth, we demonstrate that anomaly detection techniques can recall over half of the radio transients in the 10 per cent of the data with the highest anomaly scores. We find that the choice of anomaly detection algorithm makes a minor difference, but that feature set choice is crucial, especially when considering available resources for human inspection and/or follow-up. Active learning, where human labels are given for just 2 per cent of the data, improves recall by up to 20 percentage points, depending on the combination of features and model used. The best-performing results produce a factor of 5 times fewer sources requiring vetting by experts. This is the first effort to apply anomaly detection techniques to finding radio transients and shows great promise for application to other data sets, and as a real-time transient detection system for upcoming large surveys.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".