A robust method for identifying Be stars in the LAMOST Data Release 11 based on Deep-learning approach

TL;DR

This work tackles the scalable identification of Be stars in large spectroscopic surveys by first training a BiLSTM-CNN classifier to reliably label B-type stars in LAMOST DR11 low-resolution spectra. Be stars are then selected via cross-matching with Hα emission catalogs and refined through MKCLASS and visual verification, yielding 8,298 Be stars, including 4,511 new discoveries. Infrared color criteria using extinction-corrected AllWISE photometry distinguish Classical Be from Herbig Be stars, resulting in 3,363 CBe and 35 HBe objects. The authors release the B-type and Be catalogs along with training code to facilitate community research, demonstrating a robust, scalable approach to Be-star census construction and enabling studies of stellar evolution and disk physics.

Abstract

Be stars are rapidly rotating B-type stars that exhibit Balmer emission lines in their optical spectra. These stars play an important role in studies of stellar evolution and disk structures. In this work, we carried out a systematic search for Be stars based on LAMOST spectroscopic data. Using low-resolution spectra from LAMOST DR11, we constructed a data set and developed a classification model that combines long short-term memory networks and convolutional neural networks , achieving a testing accuracy of 97.86%. The trained model was then applied to spectra with signal-to-noise ratios greater than 10, yielding 55,667 B-type candidates. With the aid of the MKCLASS automated classification tool and manual verification, we finally confirmed 40,223 B-type spectra. By cross-matching with published Hα emission-line star catalogs, we obtained a sample of 8298 Be stars, including 3787 previously reported Be stars and 4511 newly discovered. Furthermore, by incorporating color information, we classified the Be star sample into Herbig Be stars and Classical Be stars. In total, we identified 3363 Classical Be stars and 35 Herbig Be stars. The B-type and Be star catalogs derived in this study, together with the code used for model training, have been publicly released to facilitate community research.

Figures (9)

• Figure 1: An example of cosmic-ray removal and missing-value correction. The green line represents the original spectrum, while the orange line shows the spectrum after processing.
• Figure 2: Comparison of spectra before and after processing. The upper panel shows the original spectrum, while the lower panel presents the spectrum after truncation and cubic spline continuum fitting and normalization.
• Figure 3: The structure of the BiLSTM-CNN model and the structure of the corresponding LSTM unit and CNN unit.
• Figure 4: Confusion matrix of the B-type star identification model.
• Figure 5: Illustration of H$\alpha$ emission lines. The upper panel shows a low-resolution spectrum without a clear H$\alpha$ emission feature, while the lower panel displays the corresponding medium-resolution spectrum, in which the H$\alpha$ emission line is clearly visible.