New theory shows how learning algorithms can create a bias towards negative outcomes.
problem Negativity bias in adaptive learning algorithms.
method Generalization of the Hot Stove Effect to settings with negative estimates leading to smaller sample sizes.
result Negativity bias persists even when negative estimates do not lead to avoidance.
Paper connects sampling and labeling biases in large-output spaces.
problem Efficient training in large-output spaces with label imbalance.
method Unified approach to address sampling and labeling biases.
result Different negative sampling schemes trade-off performance on dominant and rare labels.
From scientific experiments to online A/B testing, the previously observed data often affects how future experiments are performed, which in turn affects which data will be collected. Such adaptivity introduces complex correlations between the data and the collection procedure. In this paper, we prove that when the dat…
The paper corrects bias in synthetic data for imbalanced learning.
problem Challenges in balancing false positive and negative rates in imbalanced data.
method Proposes a bias correction procedure to generate synthetic data for minority groups.
result Enhances prediction accuracy while avoiding overfitting.
Bayesian active learning tackles nuisance parameters, leading to bias and dilemmas.
problem Bayesian active learning with nuisance parameters leads to bias and dilemmas.
method Characterizes and mitigates negative interference by accurately estimating nuisance parameters.
result The extent of negative interference can be extremely large, and accurate estimation of nuisance parameters is critical.
Depth uncertainty networks don't improve with bias correction, contrary to expectations.
problem Improving performance in active learning with overparameterised models like NNs.
method Depth uncertainty networks, compared to underparameterised models, show no improvement in performance with bias correction.
result Depth uncertainty networks do not improve with bias correction, unlike underparameterised models.
GNCL algorithm controls diversity in deep ensembles.
problem Managing bias and variance in deep ensembles.
method Generalized bias-variance decomposition for arbitrary loss functions, leading to GNCL algorithm.
result Explicit control over ensemble diversity and smooth interpolation between independent and joint training.
It is well known that in stochastic multi-armed bandits (MAB), the sample mean of an arm is typically not an unbiased estimator of its true mean. In this paper, we decouple three different sources of this selection bias: adaptive \emph{sampling} of arms, adaptive \emph{stopping} of the experiment, and adaptively \emph{…
We present a large-scale study of gender bias in occupation classification, a task where the use of machine learning may lead to negative outcomes on peoples' lives. We analyze the potential allocation harms that can result from semantic representation bias. To do so, we study the impact on occupation classification of…
New NMF method aims to improve fairness in machine learning.
problem Fairness and bias in machine learning algorithms.
method Modification of NMF objective function using min-max formulation, with two minimization methods.
result The method can sometimes improve fairness but may increase error for some individuals.
We consider the problem of estimating the parameters of a d-dimensional rectified Gaussian distribution from i.i.d. samples. A rectified Gaussian distribution is defined by passing a standard Gaussian distribution through a one-layer ReLU neural network. We give a simple algorithm to estimate the parameters (i.e., th…
Machine learning models predict brain age with systematic bias, corrected in this study.
problem Systematic bias in machine learning regression models for brain age prediction.
method General constrained optimization approach to correct bias.
result Our method effectively eliminates the bias from brain age predictions.
ADB framework improves OOD generalization by increasing ID bias during training.
problem Machine learning models degrade on new data distributions.
method ADB framework introduces controlled statistical diversity during training.
result Higher in-distribution bias leads to better out-of-distribution generalization.
Deep RL algorithms can overfit to early experiences, leading to poor performance.
problem Overfitting to early interactions in deep reinforcement learning.
method Proposed a mechanism to periodically reset part of the agent to mitigate overfitting.
result Periodic resetting improves performance in both discrete and continuous action domains.
New methods learn from PU data with non-representative positives.
problem Learning from PU data with non-representative positive classes.
method Integrates negative-unlabeled and unlabeled-unlabeled learning, or uses a recursive risk estimator.
result Effective across various real-world datasets and forms of positive bias.
This paper evaluates debiasing methods on word embeddings to reduce religious bias.
problem Social biases persist in word embeddings, potentially amplifying them in AI applications.
method Investigates and evaluates three multiclass debiasing techniques on three word embeddings.
result ConceptorDebiasing is the most effective method, reducing religious bias by 82-96%.
This paper examines fairness and arbitrariness in bias mitigation methods.
problem Understanding how different bias mitigation strategies affect individual predictions and whether they introduce arbitrariness.
method FRAME framework to evaluate bias mitigation through five dimensions: Impact Size, Change Direction, Decision Rates, Affected Subpopulations, and Neglected Subpopulations.
result Significant differences in the behaviors of debiasing methods were exhibited, highlighting the limitations of current fairness criteria and the inherent arbitrariness in the debiasing process.
Model captures asymmetric extreme events in financial returns.
problem Capturing asymmetric extreme events in financial returns.
method Two-tailed peak-over-threshold Hawkes model.
result Extreme losses contribute twice as much as gains but decay more quickly.
New method corrects risk estimation bias, improving backtesting results.
problem Underestimation of risk by existing methods, especially in small samples.
method Proposes a new algorithm for bias correction using generalized Pareto distributions.
result The new algorithm leads to improved efficiency in estimating risk with heavy tails or heteroscedasticity.
New algorithm corrects risk estimation bias for heavy-tailed data.
problem Underestimation of risk in banking and insurance due to bias in estimation procedures.
method Proposes a new algorithm for bias correction and applies it to generalized Pareto distributions.
result The algorithm leads to more accurate risk estimation, especially in heavy-tailed data.
Paper proposes MS-k-NN for improved convergence rate in k-NN classification.
problem Improving convergence rate of k-NN classification methods.
method Proposes MS-k-NN that extrapolates unweighted k-NN estimators to k=0.
result MS-k-NN achieves improved convergence rate under certain conditions.
Improved visual representation learning with conditional negative sampling.
problem Learning strong unsupervised visual representations using contrastive learning.
method Introduce a family of mutual information estimators that sample negatives conditionally.
result Improves accuracy by 2-5% points on four standard image datasets.
DCEM algorithm reduces bias in machine learning models trained on selective labels.
problem Bias in machine learning models trained on selective labels.
method Disparate Censorship Expectation-Maximization (DCEM) algorithm.
result DCEM improves bias mitigation without sacrificing discriminative performance.
Boundary effects inflate variance in Gaussian processes, leading to acquisition bias.
problem Boundary-induced acquisition bias in Gaussian processes.
method Traced root cause to geometric mechanism of kernel truncation at domain boundaries.
result Boundary effects create distortion that worsens with dimensionality, affecting acquisition behavior.
From only positive (P) and unlabeled (U) data, a binary classifier could be trained with PU learning, in which the state of the art is unbiased PU learning. However, if its model is very flexible, empirical risks on training data will go negative, and we will suffer from serious overfitting. In this paper, we propose a…
Regularized training of an autoencoder typically results in hidden unit biases that take on large negative values. We show that negative biases are a natural result of using a hidden layer whose responsibility is to both represent the input data and act as a selection mechanism that ensures sparsity of the representati…
A new differentiable divergence for time series comparison.
problem Computing discrepancies between time series of varying lengths.
method Proposed a new divergence, soft-DTW divergence, addressing issues of differentiability and positivity.
result Showed that the new divergence is a valid divergence: non-negative and minimized when time series are equal.
Cross-validation is the de facto standard for predictive model evaluation and selection. In proper use, it provides an unbiased estimate of a model's predictive performance. However, data sets often undergo various forms of data-dependent preprocessing, such as mean-centering, rescaling, dimensionality reduction, and o…
Basic binary relations such as equality and inequality are fundamental to relational data structures. Neural networks should learn such relations and generalise to new unseen data. We show in this study, however, that this generalisation fails with standard feed-forward networks on binary vectors. Even when trained wit…
The paper tackles bandit problems with biased offline data by using causal methods.
problem Improving bandit algorithms with biased offline data that includes confounding and selection biases.
method Formalizes the problem from a causal perspective, categorizes biases, and derives robust bounds for each arm.
result Causal bounds can guide the bandit agent to learn a nearly-optimal decision policy and consistently reduce asymptotic regret.
Assessing the fairness of a decision making system with respect to a protected class, such as gender or race, is challenging when class membership labels are unavailable. Probabilistic models for predicting the protected class based on observable proxies, such as surname and geolocation for race, are sometimes used to …
New method corrects bias in estimating entropic risk for better decision-making.
problem Underestimation of entropic risk when data are limited.
method Parametric bootstrap procedure to overestimate entropic risk.
result Corrected method provides better risk estimates, leading to improved decision-making.
New RLHF approach mitigates bias in aligning LLMs with human preferences.
problem Algorithmic bias in RLHF leading to preference collapse.
method Preference Matching (PM) RLHF, using PM regularizer and conditional variant.
result 29% to 41% improvement in alignment with human preferences.
Study of negative ads on social media during U.S. midterm elections.
problem Understanding the effectiveness and mechanisms of negative advertising on social media.
method Machine learning for sentiment analysis, AI image recognition, ordinal regressions.
result Negative ads are less effective than previously thought, anger is a key mechanism.
Paper introduces SLS to improve label smoothing regularization.
problem Improve generalization of neural networks by softening labels.
method Structural Label Smoothing (SLS) to mitigate bias in Bayes error rate.
result Empirical validation shows SLS outperforms traditional label smoothing.
In this paper, we study the accuracy of values aggregated over classes predicted by a classification algorithm. The problem is that the resulting aggregates (e.g., sums of a variable) are known to be biased. The bias can be large even for highly accurate classification algorithms, in particular when dealing with class-…
Paper proposes an adversarial sampling method for efficient extreme classification.
problem Training classifiers over many classes is computationally expensive.
method Adversarial sampling to draw negative samples from an adversarial model.
result Significantly reduces training time by an order of magnitude.
Proposes a method to increase diversity without sacrificing meritocracy.
problem Systemic bias in datasets affecting diversity and meritocracy.
method Optimally flipping outcome labels and training classification models simultaneously.
result The price of diversity is low and sometimes negative, enhancing diversity without significantly affecting meritocracy.
We propose shifted inner-product similarity (SIPS), which is a novel yet very simple extension of the ordinary inner-product similarity (IPS) for neural-network based graph embedding (GE). In contrast to IPS, that is limited to approximating positive-definite (PD) similarities, SIPS goes beyond the limitation by introd…
Conventional collaborative filtering techniques treat a top-n recommendations problem as a task of generating a list of the most relevant items. This formulation, however, disregards an opposite - avoiding recommendations with completely irrelevant items. Due to that bias, standard algorithms, as well as commonly used …
GWIB improves counterfactual regression by balancing latent distributions and reducing selection bias.
problem Selection bias between control and treatment groups negatively impacts counterfactual regression performance.
method GWIB uses Gromov-Wasserstein information bottleneck to maximize mutual information between covariates and outcomes while penalizing kernelized mutual information between latent representations and covariates.
result GWIB consistently outperforms state-of-the-art CFR methods in ITE estimation tasks.
ZeroS improves Transformers by adding negative weights, matching or beating softmax attention.
problem Limited performance of linear attention methods, especially in long context sequences.
method Proposes Zero-Sum Linear Attention (ZeroS) that removes the zero-order term and reweights zero-sum softmax residuals.
result ZeroS matches or exceeds standard softmax attention across various benchmarks, theoretically expanding representable functions.
New method identifies causal relationships using proxy variables in the presence of unmeasured confounders.
problem Challenges in inferring causal relationships due to unmeasured confounding.
method Develops a general nonparametric approach using a single negative control outcome (NCO) and negative control exposure (NCE).
result Establishes a new identification result and proposes a kernel-based testing procedure.
RS-PINN uses randomized smoothing to speed up high-dimensional PDE simulations without sacrificing accuracy.
problem High computational cost and bias in PINNs for high-dimensional PDEs.
method Introduces Gaussian noise for stochastic smoothing of PINNs, enabling Monte Carlo derivative approximation.
result Proposes bias correction techniques and a hybrid method to optimize the bias-variance trade-off.
We introduce a new dynamical system for sequentially observed multivariate count data. This model is based on the gamma--Poisson construction---a natural choice for count data---and relies on a novel Bayesian nonparametric prior that ties and shrinks the model parameters, thus avoiding overfitting. We present an effici…
Tree-based models biased when trained on imbalanced data, requiring new calibration methods.
problem Bias in tree-based models trained on imbalanced datasets.
method Analytical calibration of random forest models, demonstrating bias in decision trees.
result Calibrating tree-based models on imbalanced data negatively impacts predictions, especially for the minority class.
Resetting from checkpoints improves DNN training with noisy labels.
problem Latent gradient bias induced by noisy labels causes overfitting.
method Stochastic resetting applied to SGD to mitigate latent gradient bias.
result Resetting significantly improves DNN generalization performance.
This paper evaluates fairness in deep metric learning and proposes a method to reduce subgroup performance gaps.
problem The negative impact of deep metric learning representations on minority subgroup performance in downstream tasks.
method Definition of fairness in DML through inter-class, intra-class, and uniformity properties; finDML benchmark; Partial Attribute De-correlation (PARADE) method.
result Bias in DML representations propagates to downstream tasks, even with balanced training data.