SpectroTranslator: Deep-neural network algorithm for homogenising spectroscopic parameters
Bibliographic record
Abstract
Context. In modern Galactic astronomy, stellar spectroscopy plays a pivotal role in complementing large photometric and astrometric surveys and enabling deeper insights to be gained into the chemical evolution and chemo-dynamical mechanisms at play in the Milky Way and its satellites. Nonetheless, the use of different instruments and dedicated pipelines in various spectroscopic surveys can lead to differences in the derived spectroscopic parameters. Aims. Efforts to homogenise these surveys onto a common scale are essential to maximising their scientific legacy. To this aim, we developed the SPECTROTRANSLATOR, a data-driven deep neural network algorithm that converts spectroscopic parameters from the base of one survey (base A) to that of another (base B). Methods. SPECTROTRANSLATOR is comprised of two neural networks: an intrinsic network, where all the parameters play a role in computing the transformation, and an extrinsic network, where the outcome for one of the parameters depends on all the others, but not the reverse. The algorithm also includes a method to estimate the importance that the various parameters play in the conversion from base A to B. Results. To demonstrate the workings of the algorithm, we applied it to transform effective temperature, surface gravity, metallicity, [Mg/Fe], and line-of-sight velocity from the base of GALAH DR3 into the APOGEE-2 DR 17 base. We demonstrate the efficiency of the SPECTROTRANSLATOR algorithm to translate the spectroscopic parameters from one base to another, directly using parameters by the survey teams. We were able to achieve a similar performance than previous works that have performed a similar type of conversion but using the full spectrum, rather than the spectroscopic parameters. This allowed us to reduce the computational time and use the output of pipelines optimised for each survey. By combining the transformed GALAH catalogue with the APOGEE-2 catalogue, we studied the distribution of [Fe/H] and [Mg/Fe] across the Galaxy and we found that the median distribution of both quantities present a vertical asymmetry at large radii. We attribute it to the recent perturbations generated by the passage of a dwarf galaxy across the disc or by the infall of the Large Magellanic Cloud. Conclusions. Several aspects still need to be refined, such as the question of the optimal way to deal with regions of the parameter space meagrely populated by stars in the training sample. However, SPECTROTRANSLATOR has already demonstrated its capability and is poised to play a crucial role in standardising various spectroscopic surveys onto a unified framework.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".