Mitigating Annotation Burden in Active Learning with Transfer Learning and Iterative Acquisition Functions
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 situations where there is a lack of readily available annotated data, active learning is a useful tactic. To increase the model’s generalization, it entails training a model on a constrained annotation budget and repeatedly choosing the best data points for additional annotation. Active learning in deep learning usually necessitates fine-tuning subsequent deep models. But this has drawbacks, too. Specifically, it requires an initially large batch of annotated data, which is feasible when there is a constrained budget for annotation in general. We challenge this problem with a strategy inspired by transfer learning, in which only shallow classifiers are taught during active learning iterations, when an already expert model is used, it functions as a feature extractor. We also suggest a new acquisition function that takes use of active learning’s iterative character. To improve robustness, this function chooses samples based on the largest change in uncertainty between the last two models’ predictions. To ensure representation, we incorporate a diversification stage where we select samples from various regions within the categorization space. Using balanced and imbalanced datasets, our strategy is compared to competitor approaches and shows superior performance in minimizing annotation burden while preserving good model accuracy.
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 | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| 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