Web: http://arxiv.org/abs/2105.13530

May 13, 2022, 1:20 a.m. | Xi Li, David J. Miller, Zhen Xiang, George Kesidis

cs.CR updates on arXiv.org arxiv.org

Data Poisoning (DP) is an effective attack that causes trained classifiers to
misclassify their inputs. DP attacks significantly degrade a classifier's
accuracy by covertly injecting attack samples into the training set. Broadly
applicable to different classifier structures, without strong assumptions about
the attacker, an {\it unsupervised} Bayesian Information Criterion (BIC)-based
mixture model defense against "error generic" DP attacks is herein proposed
that: 1) addresses the most challenging {\it embedded} DP scenario wherein, if
DP is present, the poisoned samples are …

attacks data data poisoning defense lg model on poisoning

More from arxiv.org / cs.CR updates on arXiv.org

Senior Incident Responder

@ CipherTechs, Inc. | Remote

Data Security DevOps Engineer Senior/Intermediate

@ University of Michigan - ITS | Ann Arbor, MI

Senior Penetration Tester

@ CipherTechs, Inc. | Remote

Data Analyst

@ SkyePoint Decisions | Washington, DC

POA&M Analyst

@ SkyePoint Decisions | Washington, DC

PKI Systems Engineer

@ SkyePoint Decisions | Springfield, VA