Get Started
Home
Topics
Search
Library
Research questionHow can image-poisoning attacks remain effective across supervised and self-supervised learning without obvious visual artifacts?A perturbation that disrupts supervised learning may be less effective against self-supervised objectives. Poisoning methods therefore face a recurring trade-off between cross-paradigm effectiveness and keeping images visually acceptable.
AI
Computer Vision
Image & Video Processing
Machine Learning
Latest papersRecent research connected to this question, newest first.Leveraging Imperfect Restoration for Data Availability AttackThe source theoretically analyzes CUDA and introduces Imperfect Restoration Poisoning, comparing image-based poisoning across supervised and self-supervised settings and against representative defenses. The evidence is limited to the reported experiments; the abstract does not specify particular datasets, architectures, or threat-access assumptions.research paper · Sep 4, 2026
Related questions
How can one vision model remain robust across changing and unseen adversarial perturbation budgets?How can histopathology models withstand adversarial perturbations without erasing diagnostic tissue structures?How can black-box adversarial attacks disrupt image-based continuous-control agents with fewer environment queries?How can we reliably detect memorized patient images in medical generative models at scale?