Removing spurious features can hurt model accuracy and disproportionately affect different groups.
problem Interference from spurious features in robust model performance across different groups.
method Characterization and analysis of spurious feature removal in noiseless overparameterized linear regression.
result Removal of spurious features can decrease accuracy and disproportionately affect different groups, even in balanced datasets.
SGD quickly learns a spurious XOR feature before the signal feature, revealing learning dynamics.
problem Over-reliance on spurious correlations in neural networks trained by SGD.
method Theoretical analysis of SGD on two-layer ReLU networks trained on XOR data.
result SGD learns the spurious feature first and exponentially fast, dominating the signal feature.
CLIP models robustness to spurious features is re-evaluated using a new dataset.
problem Existing robustness tests of CLIP models may not fully reflect their performance on spurious features.
method Crafted a new dataset (CounterAnimal) to reveal CLIP models' reliance on realistic spurious features.
result CLIP models are robust to spurious features learned from their training data, not ImageNet.
Self-training avoids spurious features in domain adaptation.
problem Domain shift with large differences between source and target domains.
method Entropy minimization on unlabeled target data, initialized with a source classifier.
result Entropy minimization avoids using spurious features in large domain shifts.
The paper analyzes how deep models memorize spurious features.
problem Understanding how deep models memorize spurious features in training data.
method Characterizes spurious feature memorization via model stability and feature alignment.
result Memorization of spurious features weakens as generalization capability increases.
AFR simplifies reducing reliance on spurious features, improving model performance.
problem Reducing reliance on spurious features for out-of-distribution generalization.
method Automatic Feature Reweighting (AFR) updates the model with a weighted loss.
result AFR improves model performance on benchmarks with minimal compute.
Deep models learn spurious features correlated with target, but can still perform well.
problem Spurious correlations in feature learning.
method Empirical risk minimization and specialized group robustness training.
result Feature representations learned by ERM are competitive with specialized methods.
Online algorithm detects and removes spurious features for strong generalization.
problem Predictive models learn spurious correlations from incomplete data.
method Online continual learning algorithm that detects and removes spurious features.
result Algorithm converges to solutions with strong generalization.
The study identifies spurious correlations in high-dimensional regression and quantifies their impact.
problem Spurious correlations in high-dimensional regression models.
method Statistical characterization of spurious correlations, quantifying their amount via ridge regularization.
result The value of regularization strength that minimizes test loss is in an interval where spurious correlations increase.
FeAT improves OOD generalization by learning richer features.
problem Improving feature learning for out-of-distribution (OOD) generalization.
method Feature Augmented Training (FeAT) iteratively augments and retains features from different subsets of training data.
result FeAT effectively learns richer features, boosting OOD performance.
Last layer retraining improves robustness to spurious correlations without high computational costs.
problem Neural networks can rely on spurious features like backgrounds for predictions.
method Simple last layer retraining on large models.
result Last layer retraining matches or outperforms state-of-the-art approaches on spurious correlation benchmarks.
Study identifies and analyzes spurious correlations in data-driven models.
problem Spurious correlations in data-driven models are unreliable and hard to detect.
method Collect and analyze synthetic datasets generated from causal graphs to investigate spurious correlations.
result Patterns connecting spurious correlation hypotheses and model design choices were observed.
New rationalization method avoids spurious correlations.
problem Spurious correlations in rationalization criteria.
method Invariant rationalization using game theory constraints.
result Rationales generalize better and align with human judgments.
Regularization methods can overregulate, suppressing causal features.
problem Mitigating shortcuts in models exploiting spurious correlations.
method Analysis of regularization methods and their effects on causal features.
result Regularization can overregulate, suppressing causal features.
A method identifies domain-general features using causal graph constraints and regularization.
problem Identifying domain-general features without prior knowledge of spurious features.
method Proposes a novel regularization framework based on causal graph constraints.
result Demonstrates effectiveness in both synthetic and real-world data, outperforming state-of-the-art methods.
Algorithm removes spurious concepts from neural network representations without harming task performance.
problem Spurious correlations hinder neural network out-of-distribution generalization.
method Iterative algorithm that identifies two orthogonal subspaces in neural network representation.
result Algorithm outperforms existing methods on computer vision and natural language processing 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.
Interventional domain adaptation improves feature transferability by removing spurious correlations.
problem Improper feature transferability due to spurious correlations in domain adaptation.
method Intervention strategy using unlabeled target data to generate counterfactual features and train discriminability invariance.
result Consistent performance improvements over state-of-the-art approaches in various domain adaptation tasks.
Counterfactual data augmentations may not ensure OOD robustness if performed by a context-guessing machine.
problem Deep learning models lack out-of-distribution robustness due to reliance on spurious features.
method Theoretical analysis and demonstration of counterfactual data augmentations performed by a context-guessing machine.
result Counterfactual data augmentations by a context-guessing machine do not lead to robust OOD classifiers.
Traditional wisdom in generative modeling literature is that spurious samples that a model can generate are errors and they should be avoided. Recent research, however, has shown interest in studying or even exploiting such samples instead of eliminating them. In this paper, we ask the question whether such samples can…
Current OOD benchmarks overestimate model robustness to spurious correlations.
problem Spurious correlations degrade OOD performance, but benchmarks show the opposite.
method Analyze OOD datasets for spurious correlations and derive conditions for robustness.
result Current OOD benchmarks are misspecified and overestimate model robustness.
The paper tackles spurious correlations in machine learning models and introduces counterfactual invariance.
problem Spurious correlations in machine learning models that depend on irrelevant parts of input data.
method The paper uses causal inference to stress test models and introduces counterfactual invariance as a formal requirement.
result Counterfactual invariance is a requirement for models to be robust to irrelevant perturbations in input data.
Large learning rates enhance model robustness and compressibility.
problem Achieving robustness and resource-efficiency in machine learning models.
method Identifying and utilizing large learning rates as a facilitator for robustness and compressibility.
result Large learning rates produce desirable representation properties and compare favorably to other methods.
This thesis tackles data imperfections in ML, proposing methods to prevent discrimination and spurious feature learning.
problem Data imperfections in ML training data, leading to deployment failures.
method Fair representation learning, spurious feature detection, and data augmentation for reinforcement learning.
result Methods to make ML models more robust to data imperfections.
The paper addresses interpretability issues in EBM models by improving feature selection and reducing spurious interactions.
problem Interpretability issues in EBM models, especially spurious interactions and single feature dominance.
method Alternate Cross-feature selection, ensemble features, and model configuration alteration techniques.
result Our approach improves interpretability and predictive performance of EBM models, reducing spurious interactions and single feature dominance.
Framework uses human annotations to make models robust to spurious correlations.
problem Machine learning models fail when unmeasured variables change test distributions.
method Human annotations to augment training examples, UV-DRO objective for robustness.
result Improvements of 5-10% on digit recognition task and 1.5-5% on NYPD Police Stops analysis.
DORA analyzes deep neural networks' internal representations to detect spurious correlations.
problem Detecting spurious correlations in deep neural networks' internal representations.
method DORA uses Extreme-Activation (EA) distance measure to assess representation similarities.
result Identifies internal representations capable of detecting spurious correlations.
The paper explains how data augmentation can improve domain generalization by weakening spurious correlations.
problem Machine learning models trained with observational data fail to generalize to unseen domains due to spurious correlations.
method Developed a causal perspective to explain the success of data augmentation and derived an algorithm to select effective augmentation techniques.
result Data augmentation can be used to simulate interventional data, leading to better domain generalization.
Despite alarm over the reliance of machine learning systems on so-called spurious patterns, the term lacks coherent meaning in standard statistical frameworks. However, the language of causality offers clarity: spurious associations are due to confounding (e.g., a common cause), but not direct or indirect causal effect…
The origin of the long-range memory in the non-equilibrium systems is still an open problem as the phenomenon can be reproduced using models based on Markov processes. In these cases a notion of spurious memory is introduced. A good example of Markov processes with spurious memory is stochastic process driven by a non-…
Approximate vanishing ideal is a concept from computer algebra that studies the algebraic varieties behind perturbed data points. To capture the nonlinear structure of perturbed points, the introduction of approximation to exact vanishing ideals plays a critical role. However, such an approximation also gives rise to a…
New method ISR improves domain generalization with provable guarantees.
problem Achieving reliable performance across unseen environments.
method Invariant-feature Subspace Recovery (ISR) algorithms.
result ISR can achieve provable domain generalization with fewer training environments.
The paper formalizes criteria for non-spurious and disentangled representations using causal methods.
problem Formalizing criteria for non-spurious and disentangled representations in representation learning.
method Causal perspective, counterfactual quantities, observable consequences of causal assertions.
result Computable metrics for assessing representation learning based on observed data.
This research examines rare spurious correlations in neural networks and their impact on accuracy and privacy.
problem Rare spurious correlations in neural networks and their privacy risks.
method Introducing spurious patterns correlated with a fixed class to a few training examples, analyzing ℓ2 regularization and Gaussian noise. result Rare spurious correlations can significantly impact neural network accuracy and privacy, and specific mitigation methods can be effective.
This paper tackles confounding biases in data augmentation.
problem Mitigating spurious correlations and confounding variables in training data.
method Formal analysis and counterfactual data augmentation.
result Removing confounding biases leads to invariant features and better generalization.
Accuracy on in-distribution data correlates with out-of-distribution data when data is noisy or contains nuisance features.
problem Correlation between in-distribution and out-of-distribution accuracy in noisy or feature-rich data.
method Analyzes the impact of noise and nuisance features on model performance.
result Accuracy on in-distribution and out-of-distribution data can become negatively correlated in noisy or feature-rich data.
A self-supervised debiasing method using rank regularization mitigates spurious correlations in neural networks.
problem Spurious correlations cause biases in deep neural networks, affecting generalization.
method Spectral analysis of latent representations, rank regularization, self-supervised pretraining, debiasing of downstream tasks.
result The proposed framework significantly improves generalization performance and outperforms supervised debiasing approaches.
Mitigates spurious correlations without bias labels.
problem Spurious correlations bias model performance.
method Introduces a novel training objective and debiasing method DPR.
result DPR achieves state-of-the-art performance.
GIB improves neural network generalization by dynamically selecting task-relevant features across different sequential environments.
problem Poor generalization of deep neural networks to unseen environments.
method Proposes a gated information bottleneck (GIB) approach that dynamically drops spurious correlations and selects task-relevant features.
result GIB outperforms other IB approaches in adversarial robustness and OOD detection.
Unified framework for fairness, robustness, and distribution shifts.
problem Diverse failure modes of machine learning systems.
method Formalizes biases as violations of conditional independence and proves equivalence conditions.
result Equivalent effects of biases in different failure modes under specific conditions.
Complex-valued neural networks avoid spurious local minima.
problem Finding spurious local minima in neural networks.
method Proved no spurious local minima for shallow complex neural networks with quadratic activations.
result Complex-valued weights eliminate spurious local minima in neural networks.
A new method selects robust features for ML models using causal discovery.
problem Challenges in feature selection for ML models with limited domain knowledge.
method Multidata causal feature selection using PC1 or PCMCI algorithms.
result The method improves model performance and provides interpretable drivers.
Study on shortcuts in deep networks, revealing their layer-wise distribution and impact.
problem Understudied impact of shortcuts on feature representations in deep networks.
method Layer-wise localization through counterfactual training on clean and skewed datasets.
result Shortcuts are distributed throughout the network, not localized in specific layers.
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.
SFP prunes ID features to improve OOD generalization without domain data.
problem Improving out-of-distribution (OOD) generalization in biased models.
method Spurious Feature-targeted model Pruning (SFP) framework.
result SFP achieves optimal OOD generalization by pruning ID features.
Model patching closes subgroup performance gaps in skin cancer classification.
problem Inconsistent model performance on specific subgroups of a class.
method Two-stage framework that models subgroup features and learns semantic transformations, followed by data augmentation.
result Reductions in robust error of up to 33% relative to best baseline on benchmark datasets.
Unordered feature sets are a nonstandard data structure that traditional neural networks are incapable of addressing in a principled manner. Providing a concatenation of features in an arbitrary order may lead to the learning of spurious patterns or biases that do not actually exist. Another complication is introduced …
Mitigates biases in reward models using variational inference.
problem Spurious correlations in reward models that align large language models with human preferences.
method Formulates data-generating process, identifies non-spurious latent variables, and uses variational inference to recover them.
result Effective mitigation of spurious correlation issues, yielding more robust reward models.