Understanding the Effect of Bias in Deep Anomaly Detection

Authors: Ziyu Ye, Yuxin Chen, Haitao Zheng

IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We establish the first finite sample rates for estimating the relative scoring bias for deep anomaly detection, and empirically validate our theoretical results on both synthetic and real-world datasets. We also provide an extensive empirical study on how a biased training anomaly set affects the anomaly score function and therefore the detection performance on different anomaly classes.
Researcher Affiliation Academia Ziyu Ye , Yuxin Chen and Haitao Zheng University of Chicago {ziyuye, chenyuxin}@uchicago.edu, htzheng@cs.uchicago.edu
Pseudocode Yes Algorithm 1: Computing the anomaly detection threshold for Problem 3.2
Open Source Code No The paper does not provide a link to its source code or explicitly state that the code is open-source.
Open Datasets Yes We empirically validate our assumptions and theoretical results on both synthetic and three real-world datasets4 (Fashion-MNIST, Stat Log (Landsat Satellite), and Cellular Spectrum Misuse [Li et al., 2019]).
Dataset Splits Yes To set the threshold, we fix the target FPR to be 0.05, and vary the number of normal data in the validation set n from {100, 1K, 10K}.
Hardware Specification No The paper does not specify any hardware details (e.g., GPU/CPU models) used for running the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup No Detailed descriptions on datasets and training configurations are listed in Appendix C. (Appendix C is not provided in the paper text).