BadSAD: Clean-Label Backdoor Attacks against Deep Semi-Supervised Anomaly Detection
BadSAD: Clean-Label Backdoor Attacks against Deep Semi-Supervised Anomaly Detection
BadSAD is a clean-label backdoor attack against Deep Semi-Supervised Anomaly Detection (DeepSAD) models. It combines trigger injection in normal images with latent-space poisoning so that triggered anomalies are misclassified as normal.
Publication
Published in the 2025 IEEE International Conference on Big Data (IEEE BigData 2025).