GWSkyNet-Multi: A Machine Learning Multi-Class Classifier for LIGO-Virgo Public Alerts
Bibliographic record
Abstract
Compact object mergers which produce both detectable gravitational waves and electromagnetic emission can provide valuable insights into the neutron star equation of state, the tension in the Hubble constant, and the origin of the r-process elements. However, electromagnetic follow-up of gravitational wave sources is complicated by false positive detections, sources which do not emit light, and the transient nature of the associated electromagnetic emission. GWSkyNet-Multi is a machine learning model that attempts to resolve these issues by providing real-time predictions of the source of a gravitational wave detection. The model uses information from Open Public Alerts (OPAs) released by LIGO-Virgo within minutes of a gravitational wave detection. GWSkyNet was first introduced in Cabero et al. (2020) as a binary classifier and uses the OPA skymaps to classify sources as either astrophysical or as glitches. In this paper, we introduce GWSkyNet-Multi, a new version of GWSkyNet which further distinguishes sources as binary black hole mergers, mergers involving a neutron star, or non-astrophysical glitches in the detectors. GWSkyNet-Multi is a sequence of three one-versus-all classifiers trained using a class-balanced and physically-motivated source mass distribution. Training on this data set, we obtain test set accuracies of 94% for BBH-versus-all, 94% for NS-versus-all, and 95% for glitch-versus-all. We obtain an overall accuracy of 93% using a hierarchical classification scheme. Furthermore, we correctly identify 36 of the 40 gravitational wave detections from the first half of LIGO-Virgo's third observing run (O3a) and present predictions for O3b sources. We also briefly discuss the interpretability of our models. As gravitational wave detections increase in number and frequency, GWSkyNet-Multi will be a powerful tool for prioritizing successful electromagnetic follow-up.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.003 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.003 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.003 |
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 source (direct Gemma or distilled Codex), 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".