NeurIPS 2025 Explorer
Concepts
Authors
Glossary
johnsanterre.github.io
backdoor detection
3 papers
ICLScan: Detecting Backdoors in Black-Box Large Language Models via Targeted In-context Illumination
Lie Detector: Unified Backdoor Detection via Cross-Examination Framework
Taught Well Learned Ill: Towards Distillation-conditional Backdoor Attack