Unveiling Quasars in Gaia DR3: Multimodal Classification and Redshift Estimation of XP Sources with Deep Learning

Paper Abstract We present a multimodal deep learning framework that classifies stars, galaxies, and quasars and accurately estimates quasar redshifts using Gaia DR3's low-resolution BP/RP (XP) spectra, alongside astrometric, photometric, and CatWISE mid-infrared data. Our hierarchical classification model achieves an overall accuracy of 99.90%. Evaluating class-specific performance under a one-vs-rest framework, a Precision of 99.61% and a Recall of 99.83% are achieved for quasars. Furthermore, our spectral redshift regression model, SpecNet-Z, attains a Root Mean Square Error (RMSE) of 0.0682, a Normalized Median Absolute Deviation ($\sigma_{\mathrm{NMAD}}$) of 0.0048, and a catastrophic outlier rate of only 0.42%, demonstrating superior performance to the template-fitting approach in Gaia DR3. Applying this framework to the ~219 million sources with public XP spectra, we produce the XP-Quasars catalog, containing 123,342 high-purity quasar candidates. This catalog includes 3,763 new candidates not listed in the official GDR3 catalog. Our work provides a robust and scalable method that serves as a valuable resource for current AGN and cosmological studies and as a vital precursor for the analysis of the complete XP spectra dataset anticipated in Gaia DR4. Data Content This dataset contains four CSV files, categorized into the derived candidate catalog and the training datasets used in our study: xp_quasars.csv: The final catalog of XP-Quasar candidates. This catalog was generated by applying our multimodal deep learning framework to the complete list of Gaia DR3 sources with available XP spectra. Training Datasets: The following three files contain the labeled sources used to train and validate our hierarchical classification and redshift estimation models: star_label.csv: Labeled dataset for stars. galaxy_label.csv: Labeled dataset for galaxies. qso_label.csv: Labeled dataset for quasars.