ARTIFICIAL NEURAL NETWORKS FOR ABNORMAL WEB-TRAFFIC DETECTION.

А.Ю. Галкова

Research output: Contribution to conferenceAbstract

Abstract

We used an LSTM autoencoder to detect abnormal web-requests in HTTP requests and URLs. Implemented models were trained on normal data, and then tested on anomalies. Precision values achieved equal to 0.93 (for HTTP data), and 0.99 (for URL). Recall values are 0.99 (for HTTP data) and 1 (for URL).
Original languageEnglish
Pages78-79
StatePublished - 2020
Externally publishedYes

Keywords

  • anomaly detection
  • lstm
  • LSTM autoencoder
  • web-attack detection
  • автоэнкодер
  • веб-атаки
  • обнаружение аномалий

Fingerprint

Dive into the research topics of 'ARTIFICIAL NEURAL NETWORKS FOR ABNORMAL WEB-TRAFFIC DETECTION.'. Together they form a unique fingerprint.

Cite this