Data-Efficient Training of Pure Vision Transformers for the Task of Chest X-ray Abnormality Detection Using Knowledge Distillation
Bibliographic record
Abstract
It is generally believed that vision transformers (ViTs) require a huge amount of data to generalize well, which limits their adoption. The introduction of data-efficient algorithms such as data-efficient image transformers (DeiT) provided an opportunity to explore the application of ViTs in medical imaging, where data scarcity is a limiting factor. In this work, we investigated the possibility of using pure transformers for the task of chest x-ray abnormality detection on a small dataset. Our proposed framework is built on a DeiT structure benefiting from a teacher-student scheme for training, with a DenseNet with strong classification performance as the teacher and an adapted ViT as the student. The results show that the performance of transformers is on par with that of convolutional neural networks (CNNs). We achieved a test accuracy of 92.2% for the task of classifying chest x-ray images (normal/pneumonia/COVID-19) on a carefully selected dataset using pure transformers. The results show the capability of transformers to accompany or replace CNNs for achieving state-of-the-art in medical imaging applications. The code and models of this work are available at https://github.com/Ouantimb-Lab/DeiTCovid.
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 | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.004 | 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".