Study reveals unique gender disparities in Bangladeshi TV advertisements and talk shows.
problem Gender disparity and skin color distribution in Bangladeshi TV content.
method Computer Vision, machine learning, head pose, gender detection, skin color estimation.
result Lighter skin tones are less prevalent than darker in Bangladeshi TV, contrary to popular perception.
New research shows machine-assisted decisions can still be unfair even when the algorithm is fair.
problem Ensuring fairness in decisions made with machine-assisted human input.
method Formal model and lab experiment to analyze how machine predictions affect human decisions.
result Excluding information about protected groups from machine predictions can increase disparities.
Black women and white men have the highest income disparity in the U.S.
problem Income inequality between black women and white men in the USA
method Dynamic microeconomic model, analyzing black and white population income since 1930
result Black females and white males are poles of overall income inequality
Differential privacy reduces model accuracy more for underrepresented groups.
problem Differential privacy impacts model accuracy differently across groups.
method Training neural networks with differential privacy (DP-SGD).
result DP-SGD reduces accuracy more for underrepresented groups.
Secure methods learn fair models without revealing sensitive attributes.
problem Training fair machine learning models without exposing sensitive data.
method Secure multi-party computation to encrypt sensitive attributes.
result Outcome-based fair models can be learned, checked, or verified without revealing sensitive attributes.
The paper analyzes and corrects disparate impact in machine learning models using information theory.
problem Systematic discrimination in machine learning models based on sensitive attributes.
method Information-theoretic framework to quantify and correct disparate impact.
result Closed-form expressions for efficient correction of input distributions to achieve statistically indistinguishable output distributions.
The paper addresses fairness in machine learning by adjusting input distributions.
problem Reducing disparate impact in machine learning models over different groups.
method The approach involves learning a counterfactual distribution to adjust input variables for disadvantaged groups.
result The method can reduce disparate impact without training a new model.
Study shows explanation disparities in machine learning models are influenced by data and model properties.
problem Disparities in post-hoc machine learning explanation methods across race and gender.
method Simulations and experiments on a real-world dataset to assess challenges to explanation disparities.
result Increased covariate shift, concept shift, and omission of covariates increase explanation disparities, especially for neural network models.
Foundation models improve wage gap decomposition by capturing omitted career history factors.
problem Estimating wage disparities using incomplete career history data.
method Fine-tuning foundation models to mitigate omitted variable bias and estimate wage gaps.
result Foundation models can decompose gender wage gaps more accurately than traditional econometric methods.
Proposes a method to quantify and decompose disparity in ML models, separating exempt and non-exempt components.
problem Quantifying disparity in ML models, especially when certain features are exempted due to their critical importance.
method Information-theoretic decomposition into exempt and non-exempt components, satisfying desirable properties.
result Proposes a measure of non-exempt disparity that satisfies all desirable properties, and shows impossibility results for observational measures.
What does it mean for an algorithm to be biased? In U.S. law, unintentional bias is encoded via disparate impact, which occurs when a selection process has widely different outcomes for different groups, even as it appears to be neutral. This legal determination hinges on a definition of a protected class (ethnicity, g…
Study reduces gender bias in web data used for image recognition.
problem Gender bias in web data amplifies in machine learning models.
method Inject corpus-level constraints for calibrating structured prediction models.
result Bias amplification decreased by 47.5% and 40.5% for multilabel classification and visual semantic role labeling.
Develops algorithm to reduce real-world inequality.
problem Reduces inequality in real-world disparities.
method Impact remediation framework using social science insights and constrained optimization.
result Optimal intervention policies discovered to improve equity.
The paper examines fairness issues in decision-making systems when protected class labels are unobserved.
problem Fairness assessment challenges when protected class labels are unavailable.
method Decomposes biases in estimating outcome disparity via threshold-based imputation and proposes a weighted estimator.
result Threshold-based imputation generally overestimates disparities, while the weighted estimator has a simpler negative bias.
New method detects bias in AI models that generate data.
problem Detecting bias in AI models that generate data.
method Formalized causal fairness in generative AI, derived new decomposition results, established identification conditions, and introduced efficient estimators.
result Demonstrated the value of new methodology in analyzing bias in large language models.
Optimal pre-processing reduces disparate impact by minimizing total variation distance.
problem Achieving fairness in data outputs based on protected attributes.
method Using pre-processing to enforce fairness, minimizing total variation distance between pre-processed and original data distributions.
result The problem of fairness can be formulated as a linear program, efficiently solvable.
We analyze and develop a quantitative model describing the evolution of personal income distribution, PID, for males and females in the U.S. between 1930 and 2014. The overall microeconomic model, which we introduced ten years ago, accurately predicts the change in mean income as a function of age as well as the depend…
Paper proposes a new algorithm to minimize AUC disparities in machine learning models.
problem Minimizing unfairness in AUC scores for machine learning models.
method Proposes a minimax learning and bias mitigation framework for AUC optimization.
result Proves the convergence of the proposed algorithm to minimize group-level AUC.
Develops methods to measure and reduce fairness in datasets with limited protected attribute labels.
problem Measuring and reducing fairness in datasets with limited protected attribute labels.
method Proposes methods to estimate fairness metrics and train models to limit fairness violations using probabilistic protected attribute labels.
result Our methods provide tighter bounds on true disparity and effectively reduce fairness violations with lesser fairness-accuracy trade-offs.
Paper proposes Rényi correlation for fair machine learning.
problem Systematic discrimination in machine learning models.
method Develops a min-max formulation to balance accuracy and fairness.
result Proposes an iterative algorithm with convergence guarantees.
The paper tackles fair policy targeting by optimizing allocation rules to minimize unfairness.
problem Discrimination in individualized treatments of social welfare programs.
method Formulated as a mixed-integer linear program, solved using off-the-shelf algorithms, derived regret bounds and small sample guarantees.
result Designs fair and efficient treatment allocation rules within the Pareto frontier.
Paper proposes GN-GloVe to learn gender-neutral word embeddings.
problem Inherit strong gender stereotypes in embeddings trained on human-generated corpora.
method Proposes a novel training procedure to isolate gender information in word vectors.
result GN-GloVe successfully isolates gender information without sacrificing functionality.
Two diversity models improve subset selection for image classification tasks.
problem Data scarcity and high costs in human labeling for supervised learning.
method Facility-Location and Disparity-Min models for training data subset selection and active learning.
result Subset selection improves accuracy by 2-3% with less training data.
Develops fair ASCVD risk models reducing group disparities.
problem Inconsistent performance of risk stratification models across race and gender groups.
method Adversarial learning and large observational cohort from EHRs.
result Reduced variability in error rates across groups.
Introduces MPR to measure and optimize representation across intersectional groups in retrieval.
problem Harmful stereotypes, cultural erasure, and social disparities in image search and retrieval.
method Develops MPR metric, practical estimation methods, theoretical guarantees, and optimization algorithms.
result Optimizing MPR yields more proportional representation across multiple intersectional groups, often with minimal retrieval accuracy compromise.
Study uses deep learning to predict gender and analyze HPV vaccine perceptions on Twitter.
problem Analyzing gender differences in public perceptions on HPV vaccine using social media data.
method Convolutional neural network model trained on Twitter text for gender prediction, then applied to HPV vaccine related tweets.
result Identified gender differences in public perceptions on HPV vaccine, consistent with previous studies.
Study shows gender bias in occupation classification tasks.
problem Gender bias in machine learning for occupation classification.
method Analyzed impact of explicit gender indicators in semantic representations of biographies.
result True positive rates differ between genders, correlating with existing gender imbalances.
Deep learning improves gender classification from handwriting.
problem Classifying gender from handwritten text.
method Convolutional Neural Network (CNN) for feature extraction and gender classification.
result Deep learning approach outperforms human examiners in gender classification accuracy.
Dynamic topic model improves mental health note analysis for children.
problem Lack of longitudinal topic models for psychiatric clinical notes.
method Developed a dynamic topic model with consistent topics and individualized temporal dependencies.
result Achieved a 38% increase in topic coherence.
New research shows transparent treatment disparity is better for achieving impact parity.
problem Achieving impact parity in ML models when group membership is correlated with other features.
method Theoretical analysis and experimental testing of disparate learning processes (DLPs).
result DLPs can lead to unintended treatment disparity and within-class discrimination, undermining impact parity.
Study evaluates gender bias in relation extraction systems.
problem Gender bias in relation extraction systems.
method Created WikiGenderBias dataset, evaluated systems for bias, analyzed bias mitigation techniques.
result NRE systems exhibit gender bias in predictions.
Improved neural model predicts gender from tweets.
problem Predicting gender from Twitter text.
method RNN model with attention, LSA-reduced n-gram features.
result Improved model achieves state-of-the-art performance on English tweets.
Reduces gender classification bias by learning race-invariant face representations.
problem Societal bias in gender recognition systems.
method Adversarially trained autoencoder model to learn race-invariant face representations.
result Achieved a significant drop of over 40% in racial bias surrogate metric with race invariant representations.
The paper studies and mitigates accuracy disparity in regression models.
problem Accuracy disparity between different demographic subgroups in high-stakes domains.
method Error decomposition theorem and distribution alignment algorithm.
result The proposed algorithm effectively mitigates accuracy disparity while maintaining predictive power.
Study predicts gender from brain FC at multiple scales using deep learning and Bayesian methods.
problem Predicting gender from brain functional connectivity.
method Deep learning and Bayesian deep learning applied to brain FC data from 1003 healthy adults.
result Bayesian deep learning provides accurate predictions and uncertainty information.
Debiasing techniques can worsen gender bias in text classification, but a tweak improves both.
problem Debiasing techniques can inadvertently increase gender bias in text classification.
method Investigated traditional debiasing techniques and found they worsen bias. Suggested a minor adjustment.
result A minor adjustment to debiasing techniques can reduce gender bias while maintaining high classification accuracy.
Study quantifies gender bias in language models across 7 languages.
problem Measuring gender bias in language models across multiple languages.
method Curated dataset of politicians, multilingual language models, probing language models.
result Larger language models do not show significant gender bias compared to smaller ones.
The paper addresses fairness in dynamic pricing for strategic buyers.
problem Price disparities among specific groups can lead to unfair perceptions and legal violations.
method Proposes a dynamic pricing policy that achieves fairness and discourages strategic behavior.
result Achieves an upper bound of O ( T + H ( T ) ) O(\sqrt{T}+H(T)) O ( T + H ( T )) regret over T T T time horizons, reducing regret by 35.06% compared to a benchmark policy. Study measures gender bias in machine translation using multiple reference points.
problem Measuring and identifying gender bias in machine translation.
method Used an optimal non-biased translator, reference points from occupational statistics and survey.
result Found bias against both genders, but more against women, and found occupations have a greater effect than adjectives.
The blind application of machine learning runs the risk of amplifying biases present in data. Such a danger is facing us with word embedding, a popular framework to represent text data as vectors which has been used in many machine learning and natural language processing tasks. We show that even word embeddings traine…
New method neutralizes gender bias in word embeddings without losing semantic information.
problem Gender biases in word embeddings trained on human-generated corpora.
method Latent Disentanglement and Counterfactual Generation with siamese auto-encoder and gradient reversal layer.
result Our method outperforms existing debiasing methods in preserving semantic information and neutralizing gender biases.
Study shows unequal success of membership inference attacks across different groups.
problem Unequal success of membership inference attacks across different groups.
method Established conditions for preventing MIAs and derived connections to fairness and differential privacy.
result Estimating disparate vulnerability to MIAs can lead to overestimation; suitable attacks and statistical framework provided.
The paper offers methods for testing disparate impact using confidence intervals.
problem Testing for disparate impact in machine learning.
method Asymptotic distribution of indexes and use of confidence intervals.
result The importance of using confidence intervals over single values in testing disparate impact.
Develops methods for fair classification under linear disparity constraints.
problem Disparate impacts of machine learning algorithms on protected groups.
method Bayes-optimal fair classification methods via pre-, in-, and post-processing.
result Explicit forms of Bayes-optimal fair classifiers under linear disparity measures.
This study examines gender bias in Dutch newspapers from 1950-1990 using word embeddings.
problem Examining gender bias in historical newspapers.
method Word embeddings to measure bias changes over time.
result Clear differences in gender bias and changes within newspapers over time.
Study decomposes racial healthcare disparities via shifts in mediator distributions.
problem Racial disparities in healthcare expenditures and their underlying drivers.
method Framework decomposing disparities into mediator distribution shifts and residual components, using MEPS data.
result Substantial disparities persist even when mediators are equalized, suggesting unmeasured or structural factors.
Machine learning improves risk assessment for gender-based violence victims.
problem Accurately predicting recidivism risk in gender-based crime victims.
method Applied machine learning techniques to create models predicting recidivism risk.
result Proposed ML method outperforms classical statistical methods.
Paper improves gender detection on social media using deep learning.
problem Traditional classifiers struggle with social media data volume.
method Ensemble deep learning with multi-model architectures.
result Improved gender detection accuracy on social media posts.