AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

A Separation Result Between Data-oblivious and Data-aware Poisoning Attacks

Samuel Deng, Sanjam Garg, Somesh Jha, Saeed Mahloujifar, Mohammad Mahmoody, Abhradeep Guha Thakurta

NeurIPS 2021 · Adversarial robustness and security › Attacks on text and graphs · approach: Clean-label data poisoning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Give the attacker access to the clean training set before crafting poison examplesLASSO feature selection · Gaussian feature selectionprobability of changing selected feature support: at most ε1 vs. at least ε2Improved
Increase the fraction of the training data visible to the attackerLASSO feature-selection poisoning · Gaussian synthetic datanumber of poison points needed to add a featureImproved
Increase the fraction of the training data visible to the attackerLASSO feature-selection poisoning · SMK and TOXnumber of poison points needed to add a featureMixed
Increase the fraction of the training data visible to the attackerLASSO feature-selection poisoning · MNISTnumber of poison points needed to add a featureImproved
Use a data-aware attack that targets a rotated halfspace selected using the training setERM · Gaussian halfspace classificationtest errorHurt
Use the tested oblivious poisoning strategies instead of selecting poison examples from the training setERM · Gaussian halfspace classificationtest errorNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy