Machine learning techniques for galaxy imagery and photometry
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
In the past two decades, autonomous digital sky surveys have enabled significant advances in astronomy by collecting massive databases of imagery and other information. The quantity of data, coupled with the variety of scientific questions that require its analysis, makes manual analysis of these data impractical. To address this challenge, machine learning algorithms have been widely adopted for data analysis and product generation in astronomy. In this dissertation I examine the efficacy of machine learning algorithms such as deep convolutional neural networks, support vector machines, and vision transformers for the purpose of astronomical data analysis, with emphasize on extra-galactic objects. These include algorithms that can annotate large datasets of galaxy images, and their application to premier digital sky surveys such as Pan-STARRS. Specifically, I address the following research question: How effective are machine learning algorithms for annotating astronomical data, and what are the downsides of using these algorithms for this purpose? Namely, biases that are typical to machine learning systems can influence the annotations, which may consequently lead to false conclusions when applying statistical analysis to data annotated using such systems. These biases are often difficult to identify. Overall, this research highlights the importance of careful consideration of machine learning algorithms and their potential biases when applying them to astronomical data analysis. Our findings have broad implications for the use of machine learning in astronomy and other scientific domains, as they demonstrate the importance of addressing potential biases in machine learning systems to avoid erroneous scientific conclusions.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it