A bias classifier is introduced to resist adversarial attacks.
problem Resisting adversarial attacks on deep neural networks (DNNs).
method Introducing the bias part of a DNN with Relu as the activation function as a classifier, and adding a random first-degree part to make it information-theoretically safe.
result The bias classifier is more robust than DNNs of similar size against adversarial attacks.
Ensembles improve classifier performance by reducing bias, not variance.
problem Improving classifier performance through ensemble methods.
method Extended bias-variance decomposition for classification tasks, introducing dual reparameterization.
result Ensembling reduces bias in classifiers, contrary to the traditional view.
Adversarial domain adaptation reduces sample bias in high energy physics classifier.
problem Sample bias in high energy physics classifier training.
method Adversarial domain adaptation using neural networks with gradient reversal layer.
result Successful bias removal on simulated events at the LHC.
This work addresses fairness in ML models by training and evaluating attribute classifiers under uncertain and incomplete data.
problem Challenges in fairness metrics due to uncertain and incomplete data.
method Developed a theoretical and empirical analysis to understand and improve bias estimation in the data-scarce regime.
result The test accuracy of the attribute classifier is not always correlated with its effectiveness in bias estimation.
New analysis shows bias term affects conditions for benign overfitting in linear classifiers.
problem Understanding conditions for good generalization in linear classifiers with bias terms.
method Extending Hashimoto et al.'s results to include bias terms, analyzing covariance structure and label noise.
result Benign overfitting persists in linear classifiers with bias terms, with new constraints on data's covariance structure.
We present a novel subset scan method to detect if a probabilistic binary classifier has statistically significant bias -- over or under predicting the risk -- for some subgroup, and identify the characteristics of this subgroup. This form of model checking and goodness-of-fit test provides a way to interpretably detec…
Adam's bias shifts from full-batch to max-margin of different norms for separable data.
problem Understanding Adam's implicit bias in the incremental batch setting.
method Analyzing incremental Adam on linearly separable data, constructing datasets, and using a proxy algorithm.
result Incremental Adam can converge to different max-margin classifiers depending on the dataset and batching scheme.
Unified framework for intersectionally fair AI models using MIO.
problem Bias in AI models for high-risk domains.
method Mixed-Integer Optimization (MIO) for fairness and interpretability.
result Improved performance in detecting and mitigating bias at intersections.
A new bias score method optimizes fairness in classification.
problem Ensuring fairness in binary classification under group constraints.
method Introducing bias scores and developing a post-hoc approach to adapt to fairness constraints.
result The method maintains high accuracy while ensuring fairness constraints.
Analyzes how bias evolves in SGD training across different data sub-populations.
problem Understanding bias formation during machine learning training.
method Analytical description of SGD dynamics in a teacher-student setup with Gaussian-mixture model.
result Different sub-populations influence bias at different timescales, revealing shifting classifier preferences.
Batch normalization biases linear models towards uniform margins, improving performance in binary classification.
problem Understanding the implicit bias of batch normalization in linear models and neural networks.
method Analyzing gradient descent convergence on linear models and two-layer CNNs with batch normalization.
result Gradient descent with batch normalization in linear models converges to a uniform margin classifier with an exponential convergence rate.
New method wraps black-box classifiers to reduce bias.
problem Reduction of bias in black-box predictions.
method Post-processing with α-trees and boosting algorithms.
result Demonstrated effectiveness in reducing bias across multiple fairness metrics.
(Bolukbasi et al., 2016) demonstrated that pretrained word embeddings can inherit gender bias from the data they were trained on. We investigate how this bias affects downstream classification tasks, using the case study of occupation classification (De-Arteaga et al.,2019). We show that traditional techniques for debi…
Paper proposes a method to estimate true positive proportion without knowing it.
problem Bias in binary classifier performance due to different positive item proportions.
method Maximum likelihood estimator for true proportion of positives.
result Method accurately estimates true positive proportion in data sets.
ABROCA assesses algorithmic bias, revealing skewed distributions that inflate results.
problem Detecting nuanced performance differences in classifier fairness.
method Study of ABROCA metric's statistical properties under various conditions.
result ABROCA distributions are skewed, inflating results by chance in imbalanced classes.
Adam optimizes linear classifiers with separable data.
problem Understanding Adam's implicit bias in linear logistic regression.
method Study of Adam's behavior on linearly separable data.
result Adam converges to a linear classifier with maximum ℓ∞-margin. This study simulates biases in classifiers to assess fairness.
problem Mitigating biases in predictive models to ensure fairness.
method Agent-based model (ABM) to generate synthetic datasets with controlled biases, applied to offline and online learning approaches.
result Demonstrates how biases in data affect classifier outcomes and how mitigations impact feature usage.
New method to evaluate visual explanations from neural networks.
problem Lack of consensus on measuring effectiveness of visual explanations.
method Proposed a new procedure for evaluating explanations using a range of sources.
result Demonstrated the benefit of combining different sources and the impact of bias parameters.
The increasingly common use of neural network classifiers in industrial and social applications of image analysis has allowed impressive progress these last years. Such methods are however sensitive to algorithmic bias, i.e. to an under- or an over-representation of positive predictions or to higher prediction errors i…
In critical decision-making scenarios, optimizing accuracy can lead to a biased classifier, hence past work recommends enforcing group-based fairness metrics in addition to maximizing accuracy. However, doing so exposes the classifier to another kind of bias called infra-marginality. This refers to individual-level bia…
We study the phenomenon of bias amplification in classifiers, wherein a machine learning model learns to predict classes with a greater disparity than the underlying ground truth. We demonstrate that bias amplification can arise via an inductive bias in gradient descent methods that results in the overestimation of the…
ARFF reduces spectral bias in SGD-trained neural networks.
problem Spectral bias in two-layer neural networks.
method Comparison of SGD and ARFF on spectral bias and robustness.
result ARFF yields a closer to zero spectral bias compared to SGD.
Distillation speeds up classifier training and provides insights into its success.
problem Empirical success of knowledge distillation without theoretical explanation.
method Study of linear and deep linear classifiers, proving a generalization bound.
result Three key factors for distillation success: data geometry, optimization bias, strong monotonicity.
Visualizes classification accuracy and label bias in neural nets and trees.
problem Identifying mislabeled cases in neural net and tree-based classifications.
method Silhouette plots and quasi residual plots of PAC (probability of alternative class).
result Comparison of different classifications using silhouette width and PAC plots.
The paper detects and identifies bias in data using a counterfactual approach.
problem Detecting and identifying bias in data, especially in medical image classification.
method A global explanation framework using the counterfactual approach to identify bias causing artifacts.
result Black frames significantly influence Convolutional Neural Network's prediction, changing benign to malignant.
Two new approaches improve decision-making in asset monitoring systems.
problem Sampling bias in risk-based active learning leads to poor performance later.
method Semi-supervised learning and discriminative classification models.
result Discriminative classifiers are more robust to sampling bias.
This paper introduces a novel parameter estimation method for the probability tables of Bayesian network classifiers (BNCs), using hierarchical Dirichlet processes (HDPs). The main result of this paper is to show that improved parameter estimation allows BNCs to outperform leading learning methods such as Random Forest…
Unified framework suppresses model bias in semi-supervised learning with decoupled sampling control.
problem Class imbalance in semi-supervised learning, especially with distributional mismatches.
method Unified framework SC-SSL with decoupled sampling control, explicit expansion capability, and adaptive sampling probabilities.
result Consistent and state-of-the-art performance across various benchmark datasets and distribution settings.
Unintended bias in Machine Learning can manifest as systemic differences in performance for different demographic groups, potentially compounding existing challenges to fairness in society at large. In this paper, we introduce a suite of threshold-agnostic metrics that provide a nuanced view of this unintended bias, by…
Deep learning models show bias and variance are aligned, not in trade-off.
problem The classical bias-variance trade-off in deep learning models.
method Empirical evidence and theoretical analysis of bias and variance in deep learning models.
result Squared bias is approximately equal to variance for correctly classified sample points in deep learning models.
Study reveals how neural network architectures bias their learning based on feature directions.
problem Understanding how neural network architectures bias their learning based on feature directions.
method Defined neural anisotropy directions (NADs) to encapsulate the directional inductive bias of architectures and provided an efficient method to identify them.
result NADs characterize the features used by CNNs to discriminate between different classes for the CIFAR-10 dataset.
The study examines when to trust confidence thresholding in pseudo-labelling regression.
problem Calibrated probabilities from classifiers used for pseudo-labelling need careful handling to avoid bias in downstream regression.
method Developed a diagnostic apparatus to predict and bound the bias induced by confidence thresholding, derived a closed-form expression for the attenuation bias.
result The bias can be predicted from the residual score variance V∗, motivating a structural separation between classifier features and downstream controls. Facial analysis models are increasingly used in applications that have serious impacts on people's lives, ranging from authentication to surveillance tracking. It is therefore critical to develop techniques that can reveal unintended biases in facial classifiers to help guide the ethical use of facial analysis technolo…
Whereas previous post-processing approaches for increasing the fairness of predictions of biased classifiers address only group fairness, we propose a method for increasing both individual and group fairness. Our novel framework includes an individual bias detector used to prioritize data samples in a bias mitigation a…
Proposes a three-stage debiasing framework to improve out-of-distribution accuracy.
problem Inaccurate uncertainty estimations in bias-only models damage ensemble-based debiasing performance.
method Calibrates the bias-only model to improve its uncertainty estimations, creating a three-stage ensemble-based debiasing framework.
result The three-stage debiasing framework consistently outperforms traditional methods in out-of-distribution accuracy.
Reduces gender bias in patient notes while maintaining medical classification accuracy.
problem Bias in natural language processing of patient notes.
method Identifying and removing gendered language using BERT-based classifiers, then augmenting data to maintain performance.
result Minimal degradation in health condition classification tasks with data augmentation.
Algorithm identifies and transfers unstable features to create robust classifiers.
problem Developing unbiased classifiers from input-label pairs alone.
method Contrast different data environments in source tasks to encode unstable features, then cluster target task data and minimize worst-case risk.
result Our method maintains robustness across synthetic and real-world environments.
Reduces false positives in classifying rare online platforms.
problem Challenges in accurately identifying rare online platforms with ML.
method Calibrated probabilities and ensembles to reduce bias.
result Significantly reduces false positives in rare event detection.
The paper tackles Neyman-Pearson classification control issues.
problem Neyman-Pearson classification's control constraint is hard to satisfy in finite samples.
method Developed refined learning procedures under two accuracy control strategies.
result Proposed methods achieve desired control levels in finite samples.
Datasets often contain biases which unfairly disadvantage certain groups, and classifiers trained on such datasets can inherit these biases. In this paper, we provide a mathematical formulation of how this bias can arise. We do so by assuming the existence of underlying, unknown, and unbiased labels which are overwritt…
We study the implicit bias of gradient descent methods in solving a binary classification problem over a linearly separable dataset. The classifier is described by a nonlinear ReLU model and the objective function adopts the exponential loss function. We first characterize the landscape of the loss function and show th…
Study analyzes bias interactions in multimodal models using simulation-based methods.
problem Analyzing dynamic bias interactions in multimodal models to ensure fairness and equity.
method Simulation-based heuristic approach to compute bias scores for text-only, image-only, and multimodal embeddings.
result Multimodal bias interactions can be amplification, mitigation, or neutral, with text bias often dominant.
A new method learns from synthetic data without needing real-world examples.
problem Learning robust classifiers from limited real-world data.
method A novel setting and algorithm exploiting synthetic data independence.
result Robust classifiers trained on synthetic data generalize well to real-world domains.
A learned generative model often produces biased statistics relative to the underlying data distribution. A standard technique to correct this bias is importance sampling, where samples from the model are weighted by the likelihood ratio under model and true distributions. When the likelihood ratio is unknown, it can b…
XEM improves multivariate time series classification with explainable models.
problem Multivariate time series classification challenges.
method Hybrid ensemble method combining explicit boosting-bagging and implicit divide-and-conquer.
result XEM outperforms state-of-the-art MTS classifiers on public datasets.
Paper characterizes fairness vs. accuracy tradeoff in classification.
problem Mitigating bias in machine learning models.
method Characterizes the tradeoff between fairness and accuracy, provides a post-processing algorithm.
result Post-processing algorithm yields optimal fair classifier when score is Bayes optimal.
Adversarial training is a principled approach for training robust neural networks. Despite of tremendous successes in practice, its theoretical properties still remain largely unexplored. In this paper, we provide new theoretical insights of gradient descent based adversarial training by studying its computational prop…
Understanding the inductive bias of neural networks is critical to explaining their ability to generalise. Here, for one of the simplest neural networks -- a single-layer perceptron with n input neurons, one output neuron, and no threshold bias term -- we prove that upon random initialisation of weights, the a priori p…