A review of novelty detection

By Marco A. F. Pimentel, David A. Clifton, Lei A. Clifton, Lionel Tarassenko. 2014

1,664 citations recorded by the source corpus.

Novelty detection is the task of classifying test data that differ in some respect from the data that are available during training. This may be seen as ''one-class classification'', in which a model is constructed to describe ''normal'' training data. The novelty detection approach is typically used when the quantity of available ''abnormal'' data is insufficient to construct explicit models for non-normal classes. Application includes inference in datasets from critical systems, where the quantity of available normal data is very large, such that ''normality'' may be accurately modelled. In this review we aim to provide an updated and structured investigation of novelty detection research papers that have appeared in the machine learning literature during the last decade.

Subfields: Artificial Intelligence, Computer Networks and Communications

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 Computer Networks and Communications

Citations in the corpus

Novel Class Discovery without Forgetting (2022, background): “Methods that perform out-of-distribution detection [32,38] and open-set learning [46,18] identify instances significantly different from the training data distribution as novel samples, but do not identify sub-groups within these identified instances automatically.”
Out-of-Domain Detection for Natural Language Understanding in Dialog Systems (2019, methodology): “Significant results have been achieved by conventional methods in low-dimensional spaces [19], [22], and some of these methods have also been applied to NLU systems [23], [24].”
Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks (2017, background): “The problem of detecting out-of-distribution examples in low-dimensional space has been well-studied in various contexts (see the survey by Pimentel et al. (2014)).”
WILDS: A Benchmark of in-the-Wild Distribution Shifts (2020, positive): “Indeed, prior work has shown that selective prediction can improve model accuracy under distribution shifts (Pimentel et al., 2014; Hendrycks and Gimpel, 2017; Liang et al., 2018; Ovadia et al., 2019; Feng et al., 2019; Kamath et al., 2020).”
Adversarially Learned Anomaly Detection (2018, background): “A third class of methods uses fidelity of reconstruction [5] to determine whether an example is anomalous.”
A Gentle Introduction to Conformal Prediction and Distribution-Free Uncertainty Quantification (2021, methodology): “…distinguish outliers is sometimes called anomaly detection , novelty detection , or one-class classiﬁcation , and there are good out-of-the box methods for doing this; see [16] for an overview of outlier detection and [17, 18] for more on the statistical properties of conformal outlier detection.”
Local Novelty Detection in Multi-class Recognition Problems (2015, background): “Additionally, there exist overview papers about different approaches towards novelty detection [24, 25, 26, 28].”
Review of Bridge Structural Health Monitoring Aided by Big Data and Artificial Intelligence: From Condition Assessment to Damage Detection (2020, neutral): “There exist different novelty detection methods in machine learning (Pimentel et al. 2014; Tan 2007).”
Improving Unsupervised Defect Segmentation by Applying Structural Similarity to Autoencoders (2018, background): “(Pimentel et al., 2014) give a comprehensive overview of the field.”
Using Machine Learning Safely in Automotive Software: An Assessment and Adaption of Software Process Requirements in ISO 26262 (2018, background): “This can be coupled with additional measures, such as novelty detection (See [88] for a survey) at runtime to identify unforeseen situations that violate expectations and then take an emergency fall-back action.”
A Unifying Review of Deep and Shallow Anomaly Detection (2020, background): “This setting is arguably the most common setting in AD [159], [161], [165], [168].”
Unsupervised Energy-based Out-of-distribution Detection using Stiefel-Restricted Kernel Machine (2021, background): “Though many classical statistical models such as kernel density estimators, mixture models, and Principal Component Analysis (PCA) [8] have been studied extensively in the past for OOD detection (for a review, see [9]), this section reviews closely related approaches based on neural networks.”
A review of ensemble learning and data augmentation models for class imbalanced problems: combination, implementation and evaluation (2023, background): “[268] presented a review of research works on novelty detection in the past decade.”