notesum.ai

Published at November 21

A Dataset for Evaluating Online Anomaly Detection Approaches for Discrete Multivariate Time Series

cs.AI

cs.CE

eess.SY

Released Date: November 21, 2024

Authors: Lucas Correia¹, Jan-Christoph Goos², Thomas Bäck¹, Anna V. Kononova¹

Aff.: ¹Leiden University, Leiden, The Netherlands; ²Mercedes-Benz AG, Stuttgart, Germany

Arxiv: http://arxiv.org/abs/2411.13951v1

Refer to caption

Model	$F_{1}$	$P$	$R$	$\bar{\delta}\ [s]$
OmniA	$0.02\pm 0.02$	$0.21\pm 0.29$	$0.01\pm 0.12$	$\phantom{0}992.7\pm 69.9\phantom{x}$
TCN-AE	$0.02\pm 0.02$	$0.40\pm 0.41$	$0.01\pm 0.01$	$\phantom{0}990.5\pm 68.7\phantom{0}$
SISVAE	$0.01\pm 0.01$	$0.19\pm 0.19$	$0.01\pm 0.01$	$\mathbf{\phantom{0}990.4}\pm 67.7\phantom{0}$
LW-VAE	$0.01\pm 0.02$	$0.27\pm 0.41$	$0.01\pm 0.01$	$\phantom{0}992.4\pm 70.6\phantom{0}$
TSADIS	$0.00\pm 0.00$	$0.00\pm 0.00$	$0.00\pm 0.00$	$\phantom{0}995.8\pm 69.1\phantom{0}$
TeVAE	$\mathbf{0.03}\pm 0.03$	$\mathbf{0.48}\pm 0.43$	$\mathbf{0.02}\pm 0.02$	$\phantom{0}990.6\pm 70.5\phantom{0}$
OmniA	$0.24\pm 0.08$	$0.21\pm 0.13$	$0.34\pm 0.11$	$\phantom{0}808.7\pm 102.0$
TCN-AE	$0.14\pm 0.01$	$0.08\pm 0.01$	$0.44\pm 0.14$	$\phantom{0}726.1\pm 93.7\phantom{0}$
SISVAE	$0.11\pm 0.02$	$0.08\pm 0.01$	$0.20\pm 0.09$	$\phantom{0}863.2\pm 76.6\phantom{0}$
LW-VAE	$0.11\pm 0.01$	$0.07\pm 0.01$	$0.41\pm 0.15$	$\phantom{0}692.2\pm 114.1$
TSADIS	$0.11\pm 0.01$	$0.08\pm 0.01$	$0.35\pm 0.19$	$\phantom{0}808.4\pm 143.3$
TeVAE	$\mathbf{0.56}\pm 0.06$	$\mathbf{0.64}\pm 0.09$	$\mathbf{0.51}\pm 0.08$	$\mathbf{\phantom{0}651.0}\pm 83.0\phantom{7}$