English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Given the volume of data needed to train modern machine learning models,external suppliers are increasingly used. However, incorporating external dataposes data poisoning risks, wherein attackers manipulate their data to degrademodel utility or integrity. Most poisoning defenses presume access to a set ofclean data (or base set). While this assumption has been taken for granted,given the fast-growing research on stealthy poisoning attacks, a questionarises: can defenders really identify a clean subset within a contaminateddataset to support defenses?  This paper starts by examining the impact of poisoned samples on defenseswhen they are mistakenly mixed into the base set. We analyze five defenses andfind that their performance deteriorates dramatically with less than 1%poisoned points in the base set. These findings suggest that sifting out a baseset with high precision is key to these defenses' performance. Motivated bythese observations, we study how precise existing automated tools and humaninspection are at identifying clean data in the presence of data poisoning.Unfortunately, neither effort achieves the precision needed. Worse yet, many ofthe outcomes are worse than random selection.  In addition to uncovering the challenge, we propose a practicalcountermeasure, Meta-Sift. Our method is based on the insight that existingattacks' poisoned samples shifts from clean data distributions. Hence, trainingon the clean portion of a dataset and testing on the corrupted portion willresult in high prediction loss. Leveraging the insight, we formulate a bileveloptimization to identify clean data and further introduce a suite of techniquesto improve efficiency and precision. Our evaluation shows that Meta-Sift cansift a clean base set with 100% precision under a wide range of poisoningattacks. The selected base set is large enough to give rise to successfuldefenses.

How to Sift Out a Clean Data Subset in the Presence of Data Poisoning?