Poster: StreamToxWatch - Data Poisoning Detector in Distributed, Event-based Environments

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

StreamToxWatch, or ToxWatch for short, is an early-stage ensemble architecture for data poisoning detection and monitoring in online learning systems over streams. Detecting data poisoning is difficult, especially in distributed streaming systems where statistical baselines change on the fly and across the system. For that reason, ToxWatch employs a combination of input, (adversarial) conceptual drift, and model performance monitors intended to observe anomalous behaviors and phenomena across the system and to offer targeted detection signals to downstream applications.

Original languageEnglish
Title of host publicationDEBS 2023 - Proceedings of the 17th ACM International Conference on Distributed and Event-based Systems
PublisherAssociation for Computing Machinery, Inc
Pages182-184
Number of pages3
ISBN (Electronic)9798400701221
DOIs
StatePublished - Jun 27 2023
Event17th ACM International Conference on Distributed and Event-based Systems, DEBS 2023 - Neuchatel, Switzerland
Duration: Jun 27 2023Jun 30 2023

Publication series

NameDEBS 2023 - Proceedings of the 17th ACM International Conference on Distributed and Event-based Systems

Conference

Conference17th ACM International Conference on Distributed and Event-based Systems, DEBS 2023
Country/TerritorySwitzerland
CityNeuchatel
Period06/27/2306/30/23

Keywords

  • concept drift
  • data poisoning
  • event processing
  • stream processing

Fingerprint

Dive into the research topics of 'Poster: StreamToxWatch - Data Poisoning Detector in Distributed, Event-based Environments'. Together they form a unique fingerprint.

Cite this