Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors
cs.CR, cs.CV, cs.LG
Submitted: 2026-10-09
Updated: 2026-10-09
Comments: 18 pages, 10 figures, 10 tables
Code: https://github.com/ArnabDeyKabya/SATML
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain
- SoK: The Last Line of Defense: On Backdoor Defense Evaluation
- Label-Consistent Backdoor Attacks
- Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning
- Natural Backdoor Datasets
- Planting Undetectable Backdoors in Machine Learning Models
- Dataset Distillation
- On Defending Against Label Flipping Attacks on Malware Detection Systems
- Analysis of Label-Flip Poisoning Attack on Machine Learning Based Malware Detector
- Mitigating Label Flipping Attacks in Malicious URL Detectors Using Ensemble Trees
- Label Flipping Data Poisoning Attack Against Wearable Human Activity Recognition System
- AntiFLipper: A Secure and Efficient Defense Against Label-Flipping Attacks in Federated Learning
- Activation Gradient based Poisoned Sample Detection Against Backdoor Attacks
- Reliable Poisoned Sample Detection against Backdoor Attacks Enhanced by Sharpness Aware Minimization
- RAB: Provable Robustness Against Backdoor Attacks
- On Certifying Robustness against Backdoor Attacks via Randomized Smoothing
- Witches' Brew: Industrial Scale Data Poisoning via Gradient Matching
- Augmentation Backdoors
- Distilling the Knowledge in a Neural Network
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs