The paper proposes a method to detect and filter noisy or mislabeled data using pointwise mutual information.
problem Detecting and filtering noisy or mislabeled data in deep learning models.
method A mutual information-based framework quantifying statistical dependencies between inputs and labels.
result The method effectively filters low-quality samples, improving classification accuracy by up to 15%.
Study improves image classifier robustness to random p-norm corruptions.
problem Improving robustness of image classifiers to real-world imperceptible corruptions.
method Training and testing with random p-norm corruptions, evaluating robustness against different p-norms.
result Training with a combination of p-norm corruptions significantly improves robustness.
Study robust mean estimation under coordinate-level corruptions using Hamming distance.
problem Robust mean estimation under realistic coordinate-level corruptions.
method Introduce a novel Hamming distance-based measure and present information-theoretic analysis.
result Data cleaning-inspired approaches can match information theoretic bounds for robust mean estimation.
We introduce a new model of stochastic bandits with adversarial corruptions which aims to capture settings where most of the input follows a stochastic pattern but some fraction of it can be adversarially changed to trick the algorithm, e.g., click fraud, fake reviews and email spam. The goal of this model is to encour…
Over the last few years, the phenomenon of adversarial examples --- maliciously constructed inputs that fool trained machine learning models --- has captured the attention of the research community, especially when the adversary is restricted to small modifications of a correctly handled input. Less surprisingly, image…
We present a representation learning method that learns features at multiple different levels of scale. Working within the unsupervised framework of denoising autoencoders, we observe that when the input is heavily corrupted during training, the network tends to learn coarse-grained features, whereas when the input is …
Improved covariate shift handling with node-based Bayesian neural networks.
problem Improving generalization under covariate shift in neural networks.
method Introduced node-based Bayesian neural networks that learn latent noise variables to represent input corruptions.
result Node-based BNNs perform well under covariate shift due to input perturbations, improving uncertainty estimation and robustness.
Meta-trained optimizers improve model robustness to image corruptions.
problem Robustness of deep learning models to input noise.
method Meta-training a learned optimizer to produce robust models.
result Meta-trained optimizers improve model robustness to Gaussian noise.
New metric measures deep learning robustness to noise.
problem Deep learning models are vulnerable to noise.
method Formal definition of robustness as localized Lipschitz constant.
result New metric evaluated on competitive vision datasets.
RGI improves robustness of GAN-inversion for image restoration and anomaly detection.
problem Robustness of GAN-inversion to unknown gross corruptions.
method Proposes RGI and R-RGI methods with provable robustness guarantees.
result Restored images and corrupted region masks converge to ground truth under mild assumptions.
This paper aims to develop a new and robust approach to feature representation. Motivated by the success of Auto-Encoders, we first theoretical summarize the general properties of all algorithms that are based on traditional Auto-Encoders: 1) The reconstruction error of the input can not be lower than a lower bound, wh…
NoisyMix boosts model robustness to common corruptions.
problem Improving robustness of neural networks in real-world applications.
method NoisyMix training scheme that uses noisy augmentations in input and feature space.
result NoisyMix produces more robust models with well-calibrated class membership probabilities.
Class selectivity affects robustness to corruptions but not to adversarial attacks.
problem Understanding the relationship between class selectivity and robustness in neural networks.
method Investigated the impact of class selectivity on robustness to natural corruptions and adversarial attacks using Tiny ImageNetC and CIFAR10C datasets.
result Decreasing class selectivity increases robustness to both natural corruptions and adversarial attacks.
Self-supervised learning enhances model robustness and uncertainty.
problem Improving model robustness and uncertainty estimation.
method Self-supervised learning without requiring labels.
result Self-supervised learning improves robustness to adversarial examples, label corruption, and input corruptions.
Unsupervised learning is of growing interest because it unlocks the potential held in vast amounts of unlabelled data to learn useful representations for inference. Autoencoders, a form of generative model, may be trained by learning to reconstruct unlabelled input data from a latent representation space. More robust r…
Jacobian regularization boosts neural network robustness without degrading generalization.
problem Ensuring robustness of machine learning models against input perturbations.
method Developed a computationally efficient Jacobian regularization technique.
result Significant improvements in robustness measured against random and adversarial perturbations.
Study robust learning of Lipschitz functions under corrupted binary signals.
problem Learning a Lipschitz function with corrupted binary signals in a context of unknown corruption rounds.
method Introduced agnostic checking and new analysis techniques to design algorithms for symmetric and pricing losses.
result Achieved small cumulative loss for both symmetric and pricing losses.
Boosting framework improves adversarial robustness in deep learning.
problem Adversarial robustness of deep neural networks.
method Multiclass boosting framework with theoretical guarantees.
result Multiclass boosting achieves adversarial robustness faster than state-of-the-art methods.
Evolutionary algorithm finds optimal pixel perturbations to improve neural network generalization.
problem Minimal data corruption by pixel modifications causes overfitting in neural networks.
method Evolutionary algorithm with a novel cost function to maximize generalization gap and domain divergence.
result Method outperforms previous pixel-based data distribution shift methods on CNNs.
We give the first polynomial-time algorithm for performing linear or polynomial regression resilient to adversarial corruptions in both examples and labels. Given a sufficiently large (polynomial-size) training set drawn i.i.d. from distribution D and subsequently corrupted on some fraction of points, our algorithm out…
This paper improves test-time adaptation for distribution shifts using confidence maximization and input transformation.
problem Improving deep networks' performance on data shifted from the training distribution.
method Proposes a novel loss function combining confidence maximization and batch-wise entropy maximization with an input transformation module.
result Significantly improves robustness of pretrained networks to corruptions on benchmarks like ImageNet-C.
The paper addresses uncertainties in spectral clustering of corrupted data.
problem Uncertainties in spectral clustering due to measurement errors and missing data.
method Mathematical framework based on random set theory for Monte Carlo approximation of expected clusterings.
result Consistent quantities of interest for evaluating clusterings in corrupted data.
New approach improves model robustness and calibration in latent space.
problem Improving model robustness and calibration under input perturbations.
method VarMixup (Variational Mixup) in latent space of VAEs.
result Models trained with VarMixup in latent space are more robust and calibrated.
New algorithm learns Bayesian networks robustly in nearly-linear time.
problem Learning Bayesian networks with adversarially corrupted samples.
method Developed a nearly-linear time algorithm connecting robust learning to robust mean estimation.
result First nearly-linear time algorithm with comparable error guarantees.
A robust loss for anomaly mitigation and unsupervised contamination classification
problem Detecting and mitigating contamination in supervised and unsupervised settings
method Neural Bayesian Anomaly Mitigation (NBAM)
result Recovering the structure of contamination and identifying label-flip pairs
This paper considers the problem of matrix completion when some number of the columns are completely and arbitrarily corrupted, potentially by a malicious adversary. It is well-known that standard algorithms for matrix completion can return arbitrarily poor results, if even a single column is corrupted. One direct appl…
Develops a comprehensive theory of corruption in supervised learning.
problem Widespread corruption in data collection affects supervised learning problems.
method Introduces a general theory of corruption using Markov kernels, distinguishing and comparing corruption types.
result Establishes a unified framework for corruption types and develops mitigation strategies.
COLUMBUS discovers new features to improve domain generalization.
problem Improving machine learning models' ability to generalize to unseen domains.
method COLUMBUS uses targeted corruption of input and multi-level representations to discover new features.
result COLUMBUS achieves state-of-the-art performance on DG benchmarks.
Simple AMP algorithm robust to adversarial corruption.
problem Robust approximate message passing in spiked matrix models.
method Spectral pre-processing combined with robust spectral initialization.
result AMP output is close to correct for corrupted data.
DRO-Augment framework enhances deep neural network robustness.
problem Robustness of deep neural networks against various perturbations and adversarial attacks.
method Integrates Wasserstein Distributionally Robust Optimization with data augmentation.
result Significantly improves robustness across various corruptions and adversarial attacks.
This work studies PAC learning under evasion attacks, proving exponential sample complexity for high-dimensional inputs.
problem Formal study of PAC learning under evasion attacks where the adversary misclassifies perturbed samples.
method Proves exponential sample complexity for high-dimensional inputs under evasion attacks, formalizes hybrid attacks.
result PAC learning requires exponential sample complexity for high-dimensional inputs under evasion attacks.
Robust method estimates state, input, and parameters of linear systems online.
problem Joint estimation of state, input, and parameters in noisy or outlier-prone measurements.
method Combines recursive, alternating, and iteratively-reweighted least squares into a single algorithm.
result Good performance in presence of outliers and compared to state-of-the-art methods.
New algorithm robust to label corruptions in active learning.
problem Active learning under unknown adversarial label corruptions.
method Proposed a new active learning algorithm that is provably correct without assumptions on corruptions.
result Achieves minimax label complexity in non-corrupted setting and only requires additional labels to achieve desired accuracy in corrupted setting.
Patch Gaussian augmentation improves model robustness without sacrificing accuracy.
problem Challenges in building robust models without sacrificing accuracy.
method Adds Gaussian noise to randomly selected patches in images.
result Achieves state-of-the-art performance on benchmarks while improving clean data accuracy.
Study finds corruption negatively impacts firm performance.
problem The impact of corruption on firm performance is examined.
method Cross-sectional data analysis of a large international dataset.
result Corruption negatively affects corporate performance.
Robust testing of sparse signals in corrupted data.
problem Testing the norm of high-dimensional sparse signals in the presence of arbitrary corruption.
method Two observation models: i.i.d. samples from N(θ,Id) and sparse linear regression model. result The robust testing requires significantly more samples than non-robust testing.
New algorithm for linear optimization with adaptive corruption.
problem Stochastic linear optimization under adversarial corruption.
method Algorithm uses Löwner-John's ellipsoid for exploration and divides time into epochs.
result Regret increases linearly with corruption amount.
Unified framework for corruption-robust linear bandits with optimal gap-dependent misspecification bounds.
problem Effective learning in linear bandits with corrupted rewards across different corruption models.
method Unified framework for analyzing strong and weak corruption, connection to gap-dependent misspecification, and specialized algorithm.
result Optimal bounds for gap-dependent misspecification in linear bandits.
Sensor fusion has wide applications in many domains including health care and autonomous systems. While the advent of deep learning has enabled promising multi-modal fusion of high-level features and end-to-end sensor fusion solutions, existing deep learning based sensor fusion techniques including deep gating architec…
Detects spiky corruption in CRMDPs to learn optimal policies.
problem Learning optimal policies in environments with imperfect reward functions.
method Characterized spiky reward corruption, introduced algorithm to detect corrupt states.
result Algorithm can detect corrupt states and learn optimal policies.
We report quantitative relations between corruption level and economic factors, such as country wealth and foreign investment per capita, which are characterized by a power law spanning multiple scales of wealth and investments per capita. These relations hold for diverse countries, and also remain stable over differen…
CUTS removes corruption from models without clean data, improving utility and security.
problem Removing corruption from models without access to clean training data.
method CUTS uses a proxy set to amplify corruption and subtract it from model weights.
result CUTS recovers a large fraction of lost utility and nearly eliminates attacks with minimal damage.
Study robust estimation under varying corruption probabilities in data.
problem Robust estimation in scenarios with heterogeneous corruption rates.
method Developed estimators for mean and regression under various corruption patterns.
result Optimal estimators can discard corrupted samples beyond a specific threshold.
Network science reveals corruption risk in EU procurement markets.
problem Identifying corruption risk in EU procurement markets.
method Analyzing a large dataset of public procurement contracts using network science.
result Corruption risk is clustered and varies by country, not just by market core or periphery.
Binary classification improves with a small fraction of corrupted labels.
problem Binary classification with corrupted labels.
method Established corruption as a form of regularization and computed upper bounds on estimation error.
result Corruption is beneficial only up to a small fraction of the total sample, scaling with the square root of the sample size.
RINS-T solves time series inverse problems robustly without pretraining.
problem Recovering original signals from corrupted time series data.
method Implicit neural solvers with robust optimization techniques.
result RINS-T achieves high recovery performance without pretraining.
Student-teacher learning improves generalization with noisy inputs.
problem Transfer knowledge from clean inputs to noisy inputs.
method Analyzes student-teacher learning using deep linear networks and experiments with nonlinear networks.
result Three factors are vital for success: zero training loss, teacher knowledge, and feature decomposition.
We study the problem of corrupted sensing, a generalization of compressed sensing in which one aims to recover a signal from a collection of corrupted or unreliable measurements. While an arbitrary signal cannot be recovered in the face of arbitrary corruption, tractable recovery is possible when both signal and corrup…