The Atacama Cosmology Telescope: Machine-learning-driven Tools for Detecting Millimeter Sources in Timestream Preprocessing
Bibliographic record
Abstract
Abstract We present a new pipeline utilizing machine learning for classifying short-duration features in raw time-ordered data (TOD) of cosmic microwave background survey observations. The pipeline, specifically designed for the Atacama Cosmology Telescope, works in conjunction with the previous TOD preprocessing techniques that employ statistical thresholding to indiscriminately remove all large spikes in the data, whether they are due to noise features, cosmic rays, or true astrophysical sources, in a process called “data cuts.” This has the undesirable effect of excising real astrophysical sources, including transients, from the data. The classification pipeline demonstrated in this work uses the output from these data cuts and is able to differentiate between electronic noise, cosmic rays, and point sources, enabling the removal of undesired signals while retaining true astrophysical signals during TOD preprocessing. We achieve an overall accuracy of 90% in categorizing data spikes of different origin and, importantly, 94% for identifying those caused by astrophysical sources. Our pipeline also measures the amplitude of any detected source seen more than once and produces a subminute-to-minute light curve, providing information on its short-timescale variability. This automated pipeline for source detection and amplitude estimation will be particularly useful for upcoming surveys with large data volumes, such as the Simons Observatory.
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 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.004 |
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".