The accurate classification of broad-absorption line (BAL) quasars and non-broad-absorption line (non-BAL) quasars is key in understanding active galactic nuclei (AGN) and the evolution of the universe. With the rapid accumulation of data from large-scale spectroscopic survey projects (e.g., LAMOST, SDSS, and DESI), traditional manual classification methods face limitations. In this study, we propose a new method based on deep learning techniques to achieve an accurate distinction between BAL quasars and non-BAL quasars. We use a convolutional neural network (CNN) as the core model, in combination with various dimensionality reduction techniques, including principal component analysis (PCA), t-distributed stochastic neighborhood embedding (t-SNE), and isometric mapping (ISOMAP). These dimensionality reduction methods help extract meaningful features from high-dimensional spectral data while reducing model complexity. We employ quasar spectra from the 16th data release (DR16) of the Sloan Digital Sky Survey (SDSS) and obtain classification labels from the DR16Q quasar catalogues to train and evaluate our model. Through extensive experiments and comparisons, the combination of PCA and CNN achieve a test accuracy of 99.11%, demonstrating the effectiveness of deep learning for classifying the spectral data. Additionally, we explore other dimensionality reduction methods and machine learning models, providing valuable insights for future research in this field.

We present two complementary machine-readable catalogs for quasar (QSO) candidate selection in the Kilo-Degree Survey (KiDS) DR5. The KiDS-only catalog covers the full survey source sample and provides optical/near-infrared-based classifications together with QSO photometric redshift estimates derived without Wide-field Infrared Survey Explorer (WISE) information. The KiDS–AllWISE catalog is restricted to sources with AllWISE counterparts and provides WISE-assisted classifications and photometric redshifts. Separate stacking ensembles are trained to distinguish galaxies, stars, and QSOs. On a stratified held-out test set, the WISE-assisted classifier achieves an overall accuracy of 98.34%, a macroaveraged area under the receiver operating characteristic curve (ROC AUC) of 99.78%, and a QSO F1-score of 97.09%. Spatially disjoint R.A.-block validation yields QSO F1-scores of 95.87%–96.31%, indicating only modest spatial degradation. Feature-ablation tests show that W1/W2-based mid-infrared information provides the largest gain in QSO recovery, whereas Bayesian photometric redshift (BPZ)-derived quantities have only a minor effect. For QSO photometric redshift estimation, we train a LightGBM–XGBoost ensemble on Sloan Digital Sky Survey (SDSS) DR16 Quasar Catalog (DR16Q) quasars with valid optical and WISE photometry. On 94,057 held-out quasars, the WISE-assisted model achieves σ _NMAD = 0.0633 and a catastrophic-outlier fraction of η = 12.0%; excluding WISE features degrades these values to σ _NMAD = 0.0948 and η = 26.8%. Both catalogs provide probability-ranked QSO candidates and thresholded p _Q samples for different follow-up priorities. As the candidates are not spectroscopically confirmed, statistical applications require independent calibration of the corresponding selection functions.