Detecting Out-of-Distribution Inputs to Deep Generative Models Using Typicality

By Eric T. Nalisnick, Akihiro Matsukawa, Yee Whye Teh, Balaji Lakshminarayanan. 2019

170 citations recorded by the source corpus.

Subfields: Safety, Robustness & Interpretability

How researchers see this paper

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

In Safety, Robustness & Interpretability

Citations in the corpus

Your Classifier is Secretly an Energy Based Model and You Should Treat it Like One (2019, background): “It has been recently proposed that likelihood may not be enough for OOD detection in high dimensions (Nalisnick et al., 2019).”
A Unifying Review of Deep and Shallow Anomaly Detection (2020, background): “Another critical remark in this context is that, for (truly) high-dimensional data, the region with the highest density must not necessarily coincide with the region of highest probability mass (called the typical set), that is, the region where data points most likely occur [552].”
Task-agnostic Continual Learning with Hybrid Probabilistic Models (2021, unknown): “In particular, HCL detects task changes by measuring the typicality of the model’s statistics, which is similar to recently proposed state-of-the-art OOD detection methods by Nalisnick et al. [2019c] and Morningstar et al. [2020].”
CMG: A Class-Mixed Generation Approach to Out-of-Distribution Detection (2022, background): “Many other score functions have also been proposed, e.g., likelihood ratio [46], input complexity [50] and typicality [37].”
A Simple Fix to Mahalanobis Distance for Improving Near-OOD Detection (2021, background): “A lot of progress has been made in improving OOD detection by training complicated generative models [2; 15; 20; 14], modifying objective functions [22], and exposing to OOD samples while training [8].”
Hybrid Models for Open Set Recognition (2020, methodology): “The out-of-distribution (OOD) detection community often evaluates methods on cross-dataset setups [10, 20, 21, 26], such as training on CIFAR10 and testing on CIFAR100.”
Further Analysis of Outlier Detection with Deep Generative Models (2020, unknown): “This is also observed in [25, 19] using different generative models (autoregressive and ﬂow-based models, respectively).”
Bayesian Variational Autoencoders for Unsupervised Out-of-Distribution Detection (2019, methodology): “15 for the other experiment, and (3) the test for typicality by Nalisnick et al. (2019) ( TT ).”
Adversarial vulnerability of powerful near out-of-distribution detection (2022, background): “Improving OOD detection has seen progress by training generative models [6; 7; 2; 8], and modifying objective and loss functions [9].”
OOD Detection for EEG-based Machine Learning in High-Risk Environments (2026, unknown): “Typicality Typicality was introduced by Nalisnick et al. to correct for a failure mode of likelihoods in high dimensions where regions of high probability density can lie outside of the so-called typical set that contains almost all probability mass [7].”