Simple methods combine statistical tests for out-of-distribution detection.
problem Detecting data points not following the training distribution.
method Combining classical parametric tests (Rao's score test) and a typicality test.
result Combining Fisher's method of test statistics improves out-of-distribution detection accuracy.
In this work, we propose a method to reject out-of-distribution samples which can be adapted to any network architecture and requires no additional training data. Publicly available chest x-ray data (38,353 images) is used to train a standard ResNet-50 model to detect emphysema. Feature activations of intermediate laye…
Novel conformal methods test out-of-distribution data with labeled outliers.
problem Testing whether new data comes from the same distribution as a reference.
method Integrative conformal p-values re-weight standard p-values using dependent side information.
result The methods outperform standard conformal p-values in simulations and applications.
Study optimal ridge regularization for out-of-distribution prediction.
problem Optimal ridge regularization for predicting out-of-distribution data.
method Established conditions for optimal regularization under covariate and regression shifts, proving monotonic risk in data aspect ratio.
result Negative regularization can be optimal under shifts, even with isotropic or underparameterized training features.
New framework detects out-of-distribution samples efficiently.
problem Detecting samples from different distributions in deep neural networks.
method Statistical hypothesis testing framework combining evidence from entire network.
result Framework maintains Type I Error and achieves comparable results to state-of-the-art methods.
Paper proposes SCQ and P-TAMS for structured OOD testing in high-stakes ML.
problem Difficulty of incorporating auxiliary information in traditional conformal methods.
method Structure-adaptive conformal q-value (SCQ) and pseudo-score-guided transductive automated model selection (P-TAMS).
result Unified framework controls false discovery rate and improves power across diverse settings.
Proposes a framework for OOD detection combining multiple statistics.
problem Detecting out-of-distribution (OOD) samples reliably during inference.
method Multiple hypothesis testing with conformal p-values.
result Uniformly outperforms threshold-based tests across different datasets and neural networks.
p-DkNN uses deep representations to detect out-of-distribution data with statistical tests.
problem Lack of reliable confidence estimates in neural networks for safety-critical applications.
method Statistical testing of deep neural network's intermediate hidden representations.
result p-DkNN enables more accurate and reliable predictions by abstaining from incorrect predictions.
Research tackles ML failure in non-similar data, introducing reliable algorithms.
problem Machine learning models fail in new data distributions.
method Formal definition, assumptions, and simple algorithms for reliable generalization.
result Introduction of algorithms providing more reliable generalization.
Study on how kernel regression models generalize to out-of-distribution data.
problem Understanding generalization in machine learning models under distributional shifts.
method Replica method from statistical physics to derive analytical formula for generalization error.
result Identified overlap matrix as key determinant of generalization performance under distribution shift.
HOoD detects near-out-of-distribution groups in correlated biomedical assays.
problem Detecting near-out-of-distribution cases in biased or incomplete data.
method Projects correlated measurements through a trained model and uses permutation-based hypothesis tests.
result HOoD reliably identifies OoD groups, outperforming other detectors.
Single deep model detects out-of-distribution data with single forward pass.
problem Detecting out-of-distribution data points in neural networks.
method Deterministic uncertainty quantification (DUQ) using gradient penalty for reliable detection.
result Single model outperforms or matches ensemble methods in out-of-distribution detection.
Study compares data-driven vs model-based MRS quantification strategies, focusing on resilience to out-of-distribution effects.
problem Resilience to out-of-distribution effects in data-driven MRS quantification.
method Compared three data-driven strategies (supervised regression, self-supervised learning, test-time adaptation) against model-based fitting tools.
result Test-time adaptation proved most resilient to out-of-distribution effects, while self-supervised learning achieved intermediate performance.
A new metric predicts model performance on unseen data.
problem Predicting performance on out-of-distribution data without labels.
method Uses model predictions to pseudo-label data, trains a new model, and measures difference from in-distribution models.
result Empirically outperforms existing methods on image and text classification tasks.
Recent work has shown that deep generative models can assign higher likelihood to out-of-distribution data sets than to their training data (Nalisnick et al., 2019; Choi et al., 2019). We posit that this phenomenon is caused by a mismatch between the model's typical set and its areas of high probability density. In-dis…
Unified study of out-of-distribution generalization across 172 datasets.
problem Measuring and improving robustness of transfer learning models.
method Collect and fine-tune 31k models on 172 dataset pairs, varying architectures and settings.
result In- and out-of-distribution accuracies increase jointly but their relation is dataset-dependent.
WOOD detects out-of-distribution samples using Wasserstein distance.
problem Detecting samples from different distributions in neural networks.
method WOOD defines a Wasserstein-distance-based score to evaluate dissimilarity and solves an optimization problem.
result WOOD consistently outperforms other OOD detection methods.
Study shows gMPNNs struggle with OOD link prediction in larger test graphs.
problem Inductive out-of-distribution link prediction in larger test graphs.
method Theoretical analysis and development of a gMPNN with structural pairwise embeddings.
result Structural node embeddings from gMPNNs converge to random guessing as test graphs grow.
The paper highlights AI brittleness and the need for robust testing out-of-distribution performance.
problem The brittleness of AI systems, especially Deep Neural Networks, limits their reliability and certification.
method Analysis of AI brittleness and OOD performance, emphasizing the need for resilience and improved evaluation methods.
result AI systems are more failure-prone than certified in critical systems, and OOD performance falls off gradually.
A new method approximates pNML for faster out-of-distribution detection.
problem Detecting out-of-distribution examples efficiently.
method Influence functions approximation of pNML for neural networks.
result The approximation effectively detects out-of-distribution examples.
Recursive KalmanNet generalizes well in noisy, out-of-distribution scenarios.
problem Generalization in noisy, out-of-distribution scenarios.
method Recurrent neural network guided by a Kalman filter.
result Recursive KalmanNet performs well in scenarios with different temporal dynamics from training data.
Algorithm predicts with optimal loss by abstaining from uncertain test examples.
problem Predicting with training data not matching test data.
method Transductive abstention algorithm using labeled and unlabeled test examples.
result Optimal prediction loss guarantees with additional term for abstaining cost.
The problem of detecting whether a test sample is from in-distribution (i.e., training distribution by a classifier) or out-of-distribution sufficiently different from it arises in many real-world machine learning applications. However, the state-of-art deep neural networks are known to be highly overconfident in their…
Sharp analysis of out-of-distribution error in overparameterized models with importance weights.
problem Understanding and quantifying the degradation of performance in overparameterized models when faced with underrepresented data.
method Sharp analysis of an overparameterized Gaussian mixture model with spurious features and cost-sensitive interpolating solutions incorporating importance weights.
result Characterization of a novel tradeoff between worst-case robustness and average accuracy as a function of importance weight magnitude.
Deep learning models frequently make incorrect predictions with high confidence when presented with test examples that are not well represented in their training dataset. We propose a novel and straightforward approach to estimate prediction uncertainty in a pre-trained neural network model. Our method estimates the tr…
We describe a limitation in the expressiveness of the predictive uncertainty estimate given by mean-field variational inference (MFVI), a popular approximate inference method for Bayesian neural networks. In particular, MFVI fails to give calibrated uncertainty estimates in between separated regions of observations. Th…
Study shows different trajectory prediction models generalize better under OoD conditions.
problem Comparing trajectory prediction models' robustness across different datasets.
method Training models on Argoverse 2 and testing on Waymo Open Motion, and vice versa, with various augmentation strategies.
result Smallest model with highest inductive bias performs best in OoD generalization.
Quantum models learn unitary actions on entangled states from product states.
problem Generalization to out-of-distribution data in quantum machine learning.
method Proved out-of-distribution generalization for learning unitary actions.
result Learned unitary actions on entangled states from product states.
Improved neural networks for relational reasoning by projecting high-dimensional data to low-dimensional manifolds.
problem Out-of-distribution generalization in complex relational reasoning tasks.
method Neuroscience-inspired inductive-biased module projecting high-dimensional object representations to low-dimensional manifolds.
result Significantly better out-of-distribution generalization performance on relational reasoning tasks.
Deep Neural Networks are powerful models that attained remarkable results on a variety of tasks. These models are shown to be extremely efficient when training and test data are drawn from the same distribution. However, it is not clear how a network will act when it is fed with an out-of-distribution example. In this …
A new algorithm detects out-of-distribution samples by concentrating them in feature space.
problem Building safe AI systems requires effective out-of-distribution detection.
method The paper proposes a novel algorithm based on the observation that OoD samples concentrate in feature space.
result The algorithm achieves state-of-the-art performance on various OoD detection benchmarks.
SFB uses stable features to adapt unstable ones for better performance.
problem Improving classifier performance on out-of-distribution data by leveraging stable features.
method SFB learns a predictor that separates stable and unstable features, then adapts unstable predictions using stable predictions.
result SFB can learn an asymptotically-optimal predictor without test-domain labels.
The discriminative approach to classification using deep neural networks has become the de-facto standard in various fields. Complementing recent reservations about safety against adversarial examples, we show that conventional discriminative methods can easily be fooled to provide incorrect labels with very high confi…
Generative classifiers have been shown promising to detect illegal inputs including adversarial examples and out-of-distribution samples. Supervised Deep Infomax~(SDIM) is a scalable end-to-end framework to learn generative classifiers. In this paper, we propose a modification of SDIM termed SDIM-\emph{logit}. Instead …
Proposes CAL to learn causal adjacency for better spatiotemporal prediction.
problem Suboptimal performance in spatiotemporal prediction due to out-of-distribution data.
method Causal Adjacency Learning (CAL) method to discover causal relations over graphs.
result Calculated causal adjacency matrix enhances prediction performance on out-of-distribution test data.
We introduce two challenging datasets that reliably cause machine learning model performance to substantially degrade. The datasets are collected with a simple adversarial filtration technique to create datasets with limited spurious cues. Our datasets' real-world, unmodified examples transfer to various unseen models …
New method detects uncertainty in neural networks for out-of-distribution detection.
problem Detecting out-of-distribution inputs to ensure model reliability.
method Predictive topological uncertainty (pTU) based on persistent homology.
result pTU provides a statistical framework for OOD detection.
A new principle for extrapolating regression outside training data.
problem Regression extrapolation when predictions are outside training data range.
method Data-adaptive marginal transformation and simple relationship assumption.
result Progression method offers guarantees on approximation error beyond training data range.
A large body of recent work has investigated the phenomenon of evasion attacks using adversarial examples for deep learning systems, where the addition of norm-bounded perturbations to the test inputs leads to incorrect output classification. Previous work has investigated this phenomenon in closed-world systems where …
New research finds uncertainty estimation techniques fail to reliably detect abnormal medical cases.
problem Uncertainty estimation does not reliably detect out-of-distribution patients in medical tabular data.
method A series of tests on various uncertainty estimation techniques on real-world medical data.
result Almost all techniques fail to identify out-of-distribution patients, contradicting earlier findings.
Bayesian deep learning improves out-of-distribution detection but not always.
problem Improving the reliability of deep learning models in uncertain or novel data.
method Validation of likelihood-based Bayesian models for out-of-distribution detection.
result Bayesian deep learning models can marginally outperform conventional neural networks in certain conditions.
STOOD-X detects out-of-distribution samples without distributional assumptions and provides explainable visualizations.
problem Challenges in OOD detection, including restrictive assumptions, scalability issues, and lack of interpretability.
method Two-stage methodology combining statistical nonparametric test and explainability enhancements.
result Achieves competitive performance in high-dimensional and complex settings, with explainability framework enabling human oversight.
Paper provides a mathematical model for transformer ICL out-of-distribution generalization.
problem Understanding when transformer in-context learning can generalize beyond pre-training data.
method Minimal mathematical model of linear regression tasks with low-rank covariance matrices, analyzing distribution shifts as varying angles between subspaces.
result Transformers can generalize to all angle shifts if pre-training tasks are drawn from a union of subspaces, but not from a single Gaussian.
DAEDL improves EDL's OOD detection and classification performance by integrating feature space density.
problem Limited OOD detection and classification performance of EDL.
method Integrates feature space density with EDL's output and uses a novel parameterization.
result Demonstrates state-of-the-art performance across uncertainty estimation and classification tasks.
FOOD detects out-of-distribution samples quickly without needing OOD data.
problem Detecting out-of-distribution samples efficiently in neural networks.
method Extended DNN classifier with Gaussian layer and log likelihood ratio test.
result FOOD achieves state-of-the-art performance and is fast and applicable.
Study investigates OOD generalization methods for mechanics problems.
problem Real-world mechanics problems with unknown test environments and data distribution shifts.
method Investigates OOD generalization methods for regression problems in mechanics.
result OOD generalization methods perform better than traditional ML methods on mechanics-specific regression problems.
PFDL improves deep learning models' OOD generalization by decorrelating feature embeddings.
problem Out-of-distribution generalization in deep learning models.
method PFDL algorithm that optimizes feature decomposition network and image classification model.
result PFDL improves the accuracy of image classification models on OOD datasets.
Study identifies failure modes of machine learning models in out-of-distribution settings.
problem Machine learning models fail to generalize well to new, unseen data.
method Theoretical study of gradient-descent-trained linear classifiers on easy-to-learn tasks, followed by experiments on modern neural networks.
result Two failure modes of spurious correlations are uncovered: geometric and statistical.