Improving Adversarial Robustness of Few-Shot Learning with Contrastive Learning and Hypersphere Embedding
Bibliographic record
Abstract
Few-shot image classification (FSIC) is a computer vision task from the few-shot learning (FSL) category in which the model learns to classify images using only a few training samples. It has been demonstrated that even neural networks trained on large scale datasets are vulnerable to adversarial samples. This vulnerability is magnified in FSIC due to the low volume of training data. This paper proposes the use of hypersphere embedding and supervised contrastive learning to improve the adversarial robustness of representation learning-based FSIC. Contrastive learning contributes through its ability to bring together similar samples while pushing away dissimilar ones. On the other hand, hypersphere embedding has been successful in the representation learning tasks by restricting the embeddings to a hypersphere manifold. The proposed approach was evaluated on both 5-shot and 1-shot learning using two standard FSL networks and the standard Mini-ImageNet benchmark dataset. The evaluation shows that supervised contrastive training provides inherent adversarial robustness to the FSIC model while hypersphere embedding with cosine distance metrics improves the accuracy of the FSIC model and, when used in conjunction with an adversarial defense mechanism, boosts the adversarial performance.
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 | 0.002 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.002 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 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 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".