ADBench: Anomaly Detection Benchmark (2022-06-19T00:00:00.000000Z)

TL;DR

This work conducts the most comprehensive anomaly detection benchmark with 30 algorithms on 57 benchmark datasets, named ADBench, to identify meaningful insights into the role of supervision and anomaly types, and unlock future directions for researchers in algorithm selection and design.

Abstract

Given a long list of anomaly detection algorithms developed in the last few decades, how do they perform with regard to (i) varying levels of supervision, (ii) different types of anomalies, and (iii) noisy and corrupted data? In this work, we answer these key questions by conducting (to our best knowledge) the most comprehensive anomaly detection benchmark with 30 algorithms on 57 benchmark datasets, named ADBench. Our extensive experiments (98,436 in total) identify meaningful insights into the role of supervision and anomaly types, and unlock future directions for researchers in algorithm selection and design. With ADBench, researchers can easily conduct comprehensive and fair evaluations for newly proposed methods on the datasets (including our contributed ones from natural language and computer vision domains) against the existing baselines. To foster accessibility and reproducibility, we fully open-source ADBench and the corresponding results.

Authors

Yue Zhao

11 papers

Xiyang Hu

6 papers

Songqiao Han

2 papers

TL;DR

Abstract

Authors

References203 items

Toward Unsupervised Outlier Model Selection

OpenOOD: Benchmarking Generalized Out-of-Distribution Detection

Diffusion Models: A Comprehensive Survey of Methods and Applications

ADMoE: Anomaly Detection with Mixture-of-Experts from Noisy Labels

Why do tree-based models still outperform deep learning on tabular data?

Unsupervised Anomaly Detection by Robust Density Estimation

Sparx: Distributed Outlier Detection at Scale

Benchmarking Unsupervised Anomaly Detection and Localization

On Equivalence of Anomaly Detection Algorithms

Anomaly Detection by Leveraging Incomplete Anomalous Knowledge with Anomaly-Aware Bidirectional GANs

PyGOD: A Python Library for Graph Outlier Detection

Semi-supervised anomaly detection algorithm based on KL divergence (SAD-KL)

Data-Efficient and Interpretable Tabular Anomaly Detection

Latent Outlier Exposure for Anomaly Detection with Contaminated Data

Anomalib: A Deep Learning Library for Anomaly Detection

AutoML: state of the art with a focus on anomaly detection, challenges, and research directions

Uncovering the Source of Machine Bias

ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functions

Online false discovery rate control for anomaly detection in time series

When Does Contrastive Learning Preserve Adversarial Robustness from Pretraining to Finetuning?

A Unified Survey on Anomaly, Novelty, Open-Set, and Out-of-Distribution Detection: Solutions and Future Challenges

DP-SSL: Towards Robust Semi-supervised Learning with A Few Labeled Samples

Generalized Out-of-Distribution Detection: A Survey

Learned Robust PCA: A Scalable Deep Unfolding Approach for High-Dimensional Outlier Detection

Deep Neural Networks and Tabular Data: A Survey

Gen2Out: Detecting and Ranking Generalized Anomalies

A Survey on Open Set Recognition

Toward Explainable Deep Anomaly Detection

Explainable Deep Few-shot Anomaly Detection with Deviation Networks

Revisiting Deep Learning Models for Tabular Data

An Information Retrieval Approach to Building Datasets for Hate Speech Detection

A Comprehensive Survey on Graph Anomaly Detection With Deep Learning

Deep Clustering based Fair Outlier Detection

Feature Encoding With Autoencoders for Weakly Supervised Anomaly Detection

Do We Really Need to Learn Representations from In-domain Data for Outlier Detection?

SegmentMeIfYouCan: A Benchmark for Anomaly Segmentation

Beyond Outlier Detection: Outlier Interpretation by Attention-Guided Triplet Deviation Network

DATE: Detecting Anomalies in Text via Self-Supervision of Transformers

Neural Transformation Learning for Deep Anomaly Detection Beyond Images

Learning Placeholders for Open-Set Recognition

SSD: A Unified Framework for Self-Supervised Outlier Detection

Homophily outlier detection in non-IID categorical data

Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development

Learning Diverse-Structured Networks for Adversarial Robustness

Towards Fair Deep Anomaly Detection

FairOD: Fairness-aware Outlier Detection

AutoAudit: Mining Accounting and Time-Evolving Graphs

Further Analysis of Outlier Detection with Deep Generative Models

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

PANDA: Adapting Pretrained Features for Anomaly Detection and Segmentation

Energy-based Out-of-distribution Detection

MIMOSA: Multi-constraint Molecule Sampling for Molecule Optimization

Do Wider Neural Networks Really Help Adversarial Robustness?

Multi-attributed heterogeneous graph convolutional network for bot detection

A Unifying Review of Deep and Shallow Anomaly Detection

COPOD: Copula-Based Outlier Detection

TODS: An Automated Time Series Outlier Detection System

Meta-AAD: Active Anomaly Detection with Deep Reinforcement Learning

Enhancing Graph Neural Network-based Fraud Detectors against Camouflaged Fraudsters

Re-Thinking Co-Salient Object Detection

Explainable Deep One-Class Classification

Adapting k-means algorithms for outliers

Understanding Anomaly Detection with Deep Invertible Networks through Hierarchies of Distributions and Features

Adversarial Self-Supervised Contrastive Learning

Anomaly Detection with Domain Adaptation

Classification-Based Anomaly Detection for General Data

Benchmarking Unsupervised Outlier Detection with Realistic Synthetic Data

Outlier Detection Ensemble with Embedded Feature Selection

Event Outlier Detection in Continuous Time

PIDForest: Anomaly Detection via Partial Identification

Ring: Real-Time Emerging Anomaly Monitoring System Over Text Streams

CORE: Automatic Molecule Optimization Using Copy & Refine Strategy

Meta Label Correction for Noisy Label Learning

Deep Weakly-supervised Anomaly Detection

SELF: Learning to Filter Noisy Labels with Self-Ensembling

RADE: resource-efficient supervised anomaly detection using decision tree-based ensemble methods