Analyzing and Mitigating Poisoning Attacks on Flow-Based Intrusion Detection
Researchers backdoor flow-based IDS models with 0.001% poisoned data and propose mitigations.
A cs.CR paper shows latent backdoors can be planted in flow-based intrusion-detection models by poisoning as little as 0.001% of training data, preserving near-normal performance on clean inputs while achieving high attack success. The authors propose localizing trigger features with constrained perturbation analysis and detecting behavioral drift against a distilled baseline. Mitigation is automated data sanitization or neuron pruning.
- Latent backdoors were implanted by poisoning 0.001% of IDS training data.
- Poisoned models stay near-normal on clean inputs while attacks succeed.
- Trigger features are localized by constrained perturbation analysis.
- Mitigation uses data sanitization or neuron pruning against a distilled baseline.
From the original article
1 min read at arxiv.orgAlthough Artificial Intelligence (AI)-based Intrusion Detection Systems (IDSs) are increasingly deployed in operational networks, their dependence on large-scale training data introduces a critical vulnerability to stealthy poisoning attacks. We demonstrate that latent backdoors can be successfully implanted in flow-based IDS models by poisoning as little as 0.
By Florian Spelter · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.