Linked Presentation: Evading Provenance-Based ML Detectors with Adversarial System ActionsHOLMES: Efficient Distribution Testing for Secure Collaborative Learning
