Generalized Out-of-Distribution Detection: A Survey

By Jingkang Yang, Kaiyang Zhou, Yixuan Li, Ziwei Liu. 2021

1,452 citations recorded by the source corpus.

Out-of-distribution (OOD) detection is critical to ensuring the reliability and safety of machine learning systems. For instance, in autonomous driving, we would like the driving system to issue an alert and hand over the control to humans when it detects unusual scenes or objects that it has never seen during training time and cannot make a safe decision. The term, OOD detection, first emerged in 2017 and since then has received increasing attention from the research community, leading to a plethora of methods developed, ranging from classification-based to density-based to distance-based ones. Meanwhile, several other problems, including anomaly detection (AD), novelty detection (ND), open set recognition (OSR), and outlier detection (OD), are closely related to OOD detection in terms of motivation and methodology. Despite common goals, these topics develop in isolation, and their subtle differences in definition and problem setting often confuse readers and practitioners. In this survey, we first present a unified framework called generalized OOD detection, which encompasses the five aforementioned problems, i.e.,AD, ND, OSR, OOD detection, and OD. Under our framework, these five problems can be seen as special cases or sub-tasks, and are easier to distinguish. Despite comprehensive surveys of related fields, the summarization of OOD detection methods remains incomplete and requires further advancement. This paper specifically addresses the gap in recent technical developments in the field of OOD detection. It also provides a comprehensive discussion of representative methods from other sub-tasks and how they relate to and inspire the development of OOD detection methods. The survey concludes by identifying open challenges and potential research directions.

Subfields: Artificial Intelligence, Civil and Structural Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Civil and Structural Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Few-Shot Class-Incremental Learning by Sampling Multi-Phase Tasks (2022, background): “However, real-world applications often face the stream format data [7] and novel classes [8], [9], [10], [11], [12], e.”
Deep Class-Incremental Learning: A Survey (2023, background): “, few-shot [204], imbalanced [136], with noisy label [205], weak supervised [206], multi modal [207], concept drift [13], novel classes [208], [209] etc.”
Generalizing to Unseen Domains: A Survey on Domain Generalization (2021, background): “A related survey paper [5] is for OOD detection”
A Multi-Head Model for Continual Learning via Out-of-Distribution Replay (2022, methodology): “We employ Area Under the ROC Curve (AUC) to measure the performance of OOD detection as AUC is the standard metric used in OOD detection papers (Yang et al., 2021).”
Domain Generalization: A Survey (2021, background): “Studies in [2], [9], [10] suggest that deep learning models’ performance degrades significantly on OOD datasets, even with just small variations in the data generating process.”
ADBench: Anomaly Detection Benchmark (2022, background): “While ADBench focuses on the AD tasks, we note that there are some closely related problems, including out-of-distribution (OOD) detection [182, 183], novelty detection [116, 137], and open-set recognition (OSR) [51, 112].”
OpenOOD: Benchmarking Generalized Out-of-Distribution Detection (2022, methodology): “Recently, a generalized OOD detection framework [5] is proposed, which unifies similar tasks such as AD, OSR, and OOD detection.”
ViM: Out-Of-Distribution with Virtual-logit Matching (2022, background): “A comprehensive review of OOD and related topics including open set recognition, novelty detection and anomaly detection can be found in [38].”
Connecting the Dots in Trustworthy Artificial Intelligence: From AI Principles, Ethics, and Key Requirements to Responsible AI Systems and Regulation (2023, background): “In this level we can also place the generalized framework of Out-of-Distribution (OoD) detection [30], which refers to anomaly detection, novelty detection and open set recognition, the latter referring to the capability of the model to detect, characterize and incorporate new unknown patterns to it”
AI Autonomy: Self-initiated Open-world Continual Learning and Adaptation (2022, background): “Although novelty detection (Yang et al. 2021; Pang et al. 2021) and incremental or continual learning (Chen and Liu 2018; Parisi et al. 2019; Lomonaco et al. 2022) have been studied widely, little work has been done to build a SOLA system.”
Self-Initiated Open World Learning for Autonomous AI Agents (2021, background): “An recent survey of the topic can be found in (Yang et al. 2021).”
Open-World Continual Learning: Unifying Novelty Detection and Continual Learning (2023, methodology): “We employ Area Under the ROC Curve (AUC) to measure the performance of OOD detection as AUC is the standard metric used in OOD detection papers [65].”
Out-of-distribution Detection with Deep Nearest Neighbors (2022, neutral): “…& Gales, 2018; Hendrycks et al., 2019; Geifman & El-Yaniv, 2019; Hein et al., 2019; Meinke & Hein, 2019; Mohseni et al., 2020; Liu et al., 2020; Jeong & Kim, 2020; Van Amersfoort et al., 2020; Yang et al., 2021; Chen et al., 2021; Wei et al., 2022; Ming et al., 2022a; Katz-Samuels et al., 2022).”
OOD Detection for EEG-based Machine Learning in High-Risk Environments (2026, unknown): “An influential survey paper by Yang et al. even frames this exclusive focus as part of the general definition of OOD detection, although it is acknowledged that other shifts may be considered OOD from a generalisation perspective or when considering high-risk applications [33].”
Do You Know What You Are Talking About? Characterizing Query-Knowledge Relevance For Reliable Retrieval Augmented Generation (2024, background): “Our work is re-lated to the emerging research field of out-of-distribution (OoD) detection (Yang et al., 2021), which aims to detect test samples that are outside of the training data distribution.”