Robosurg: Resilience of Vision-Language Models Against Adversarial Attacks in Robotic Surgery
Bibliographic record
Abstract
The integration of artificial intelligence (AI) and visionlanguage models (VLMs) is revolutionizing robotic-assisted surgeries by enhancing decision-making and operational precision. Despite these advances, the robustness of VLMs against adversarial attacks, particularly in surgical scene understanding via Visual Question Answering (VQA), is not well understood. These models are integral to surgical accuracy but may be vulnerable to data perturbations that can compromise patient outcomes. This paper introduces RoboSurg, a framework designed to protect VLMs against such adversarial threats in surgical environments. We tailor adversarial attacks to the surgical domain, exposing significant vulnerabilities in widely used models such as SurgicalGPT, BLIP, BLIP-2, and CLIP-ViL. RoboSurg incorporates multimodal adversarial training, advanced input preprocessing, and targeted architectural modifications to enhance the strength of these models. Our experimental results demonstrate that RoboSurg effectively lowers the impact of adversarial threats while maintaining model performance under normal operating conditions. Furthermore, the first study of adversarial attacks on VLMs in a surgical context is presented in this paper, and bringing AI solutions to the forefront of safety and reliability marks a significant achievement.
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 | 0.010 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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".