Key Findings
Results that define how this survey should be used in the paper set.
- 01298 anomalies from 930K sources (0.03% canonical top-cut)
- 0273% novel (not in SIMBAD)
- 03203 SIMBAD-unmatched X-ray sources near LMC (not independently confirmed discoveries)
- 04Cross-validation stability 81.5% — highest of all Path-C surveys
- 05Fast processing (8 seconds) due to tabular data
Pipeline
| Model | 16-dim autoencoder latent + IsolationForest (100 trees) |
| Wavelength | X-ray (0.2-8 keV) |
| Cost | ~$1 |
| Runtime | ~8 seconds |
Top Anomalies
No top anomalies surfaced yet
This survey is still being processed or its top-N list has not been published.