Généralisation de domaine en vision par ordinateur : apport des modèles pré-entraînés à grande échelle
Bibliographic record
Abstract
In many machine learning applications, training and test data differ significantly, leading to what is commonly referred to as a domain shift. In industrial settings, such a shift typically arises when a model is trained on synthetic data and then deployed on real data. This discrepancy undermines the robustness of models, as their performance often drops when applied to test data. This thesis aims to design new methods to mitigate these performance losses and enhance generalization capabilities in the presence of domain shifts. The proposed approach leverages the knowledge encoded in large pretrained models, which emerged shortly before the start of this work, in order to exploit their rich representations for better handling such discrepancies. We first provide an evaluation of the effectiveness of these models on datasets from both academic and industrial contexts. We then introduce a domain adaptation method based on textual cues describing the target domain. While these two contributions focus on image classification, the final part of the thesis extends the approach to the task of object detection.
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 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 | 0.010 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.002 | 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) | 0.000 | 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, 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".