An adaptive parameter estimator for poor-quality spectral data of white dwarfs

White dwarfs represent the end stage for 97% of stars, making precise parameter measurement crucial for understanding stellar evolution. Traditional estimation methods involve fitting spectra or photometry, which require high-quality data. In recent years, machine learning has played a crucial role in processing spectral data due to its speed, automation, and accuracy. However, two common issues have been identified. First, most studies rely on data with high signal-to-noise ratios (S/N > 10), leaving many poor-quality datasets underutilized. Second, existing machine learning models, primarily based on convolutional and recurrent networks and their variants, fail to simultaneously capture both local spatial and long-range sequential spectral information. To address these challenges, we designed the estimator network (EstNet), an advanced algorithm integrating multiple techniques, including residual networks, squeeze and excitation attention, gated recurrent units, adaptive loss, and Monte Carlo dropout. A total of 5965 poor-quality white dwarf spectra (R~1800, S/N~1.17) were used for the experiments. EstNet achieved average percentage errors of 13.81% for effective temperature and 3.92% for surface gravity on the test set. These results are superior to other mainstream algorithms and consistent with the outcomes of traditional theoretical spectrum fitting methods. In the future, our algorithm will be applied to large-scale parameter estimations on data from the China Space Station Survey Telescope and the Dark Energy Spectroscopic Instrument.