Proposes bounds on bias from low-dimensional representations in CATE estimation.
problem Bias in CATE estimation due to low-dimensional representations.
method Proposes a refutation framework to estimate bounds on representation-induced confounding bias.
result Demonstrates effectiveness of refutation framework in practice.
Social bias in machine learning has drawn significant attention, with work ranging from demonstrations of bias in a multitude of applications, curating definitions of fairness for different contexts, to developing algorithms to mitigate bias. In natural language processing, gender bias has been shown to exist in contex…
Societal bias towards certain communities is a big problem that affects a lot of machine learning systems. This work aims at addressing the racial bias present in many modern gender recognition systems. We learn race invariant representations of human faces with an adversarially trained autoencoder model. We show that …
Paper tackles overestimation bias in continuous control, improving performance by 25%.
problem Overestimation bias in off-policy learning.
method Truncated Quantile Critics (TQC) combines distributional representation, truncation, and ensembling of critics.
result TQC outperforms state-of-the-art methods by 25% on the Humanoid environment.
Improves transferability of representations from source to target domains with weights and invariant representations.
problem Label shift between source and target domains in unsupervised domain adaptation.
method Integrates weights and invariant representations to bound the target risk, highlighting the role of inductive bias.
result Empirical evidence shows that weak inductive bias makes adaptation more robust.
Optimal feature transfer identified through bias-variance analysis.
problem Optimizing feature transfer in transfer learning.
method Simple linear model with fine-grained bias-variance decomposition.
result Optimal pretrained feature transform is naturally sparse.
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 method reduces bias in NLI models using ensemble adversarial training.
problem Spurious correlations between hypotheses and entailment classes in NLI datasets.
method Adversarial training with an ensemble of classifiers to reduce bias in sentence representations.
result Ensemble adversarial training produces more robust NLI models, outperforming previous methods.
New estimator corrects bias in CKA for sparsely sampled neurons.
problem Bias in CKA for sparsely sampled neurons.
method Novel estimator that corrects for input and feature sampling.
result Reliable model-to-brain alignment with sparsely sampled neurons.
This study examines representation bias in open-source Qwen models for investment decisions.
problem Representation bias in financial applications of large language models.
method Balanced round-robin prompting over 150 U.S. equities, constrained decoding, token-logit aggregation.
result Firm size and valuation increase model confidence, while risk factors decrease it.
Study reveals limitations of fair representation learning methods and cautions against their use in performance-sensitive tasks.
problem Limitations of fair representation learning methods in performance-sensitive tasks.
method Using causal reasoning, the study defines and formalizes different sources of dataset bias and examines the performance of fair representation learning under distribution shifts.
result Fundamental limitations on fair representation learning when evaluation data is drawn from the same distribution as training data.
This study investigates abrupt learning dynamics in Transformers, revealing plateau formation and internal representation collapse.
problem Abrupt learning in Transformers, particularly during the loss plateau.
method Investigates mechanisms of abrupt learning in shallow Transformers, focusing on attention maps and hidden states.
result Reveals plateau formation, internal representation collapse, and strong repetition bias in outputs.
A key element in transfer learning is representation learning; if representations can be developed that expose the relevant factors underlying the data, then new tasks and domains can be learned readily based on mappings of these salient factors. We propose that an important aim for these representations are to be unbi…
This paper provides an analysis of the tradeoff between asymptotic bias (suboptimality with unlimited data) and overfitting (additional suboptimality due to limited data) in the context of reinforcement learning with partial observability. Our theoretical analysis formally characterizes that while potentially increasin…
Contrastive learning struggles with class collapse and feature suppression, revealing bias towards simpler solutions.
problem Contrastive learning struggles with class collapse and feature suppression, especially in supervised and unsupervised settings.
method Unified theoretical framework to determine which features are learnt by CL, revealing bias towards simpler solutions.
result Bias towards simpler solutions is a key factor in class collapse and feature suppression.
New technique reduces bias in DNN models without sensitive attribute annotations.
problem Existing bias mitigation methods require instance-level annotations and do not guarantee removal of all sensitive information.
method Representation Neutralization for Fairness (RNF) debiases only the classification head of DNN models using neutralized representations.
result RNF effectively reduces discrimination of DNN models with minimal performance degradation.
Convolutional Neural Networks (CNNs) have become the state-of-the-art method to learn from image data. However, recent research shows that they may include a texture and colour bias in their representation, contrary to the intuition that they learn the shapes of the image content and to human biological learning. Thus,…
Many machine learning algorithms are trained and evaluated by splitting data from a single source into training and test sets. While such focus on in-distribution learning scenarios has led to interesting advancement, it has not been able to tell if models are relying on dataset biases as shortcuts for successful predi…
A new method uses algebraic insights to create approximately equivariant networks without complex architectures.
problem Designing equivariant neural networks with complex architectures and high computational cost.
method Imposes the group's regular representation as an inductive bias via an auxiliary loss, adding no learnable parameters.
result Matches or outperforms specialized models in several cases, even for infinite groups.
New method estimates shape distance in neural representations with limited data.
problem Measuring geometric similarity between high-dimensional network representations.
method Method-of-moments estimator with tunable bias-variance tradeoff.
result New estimator achieves lower bias than standard methods in high-dimensional settings.
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.
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.
Simplified non-contrastive learning avoids representation collapse.
problem Training failure modes in self-supervised learning.
method Hyperdimensional computing and inductive bias.
result The approach avoids representation collapses.
The paper introduces Relative Bias to quantify LLM bias systematically.
problem Quantifying bias in LLMs is challenging due to ambiguity and rapid model emergence.
method Relative Bias framework using Embedding Transformation and LLM-as-a-Judge methodologies.
result The two scoring methods show strong alignment, providing a systematic approach.
A central goal of meta-learning is to find a learning rule that enables fast adaptation across a set of tasks, by learning the appropriate inductive bias for that set. Most meta-learning algorithms try to find a \textit{global} learning rule that encodes this inductive bias. However, a global learning rule represented …
Unified framework interprets SSL models, revealing biases.
problem Interpreting black-box SSL models for transparency.
method KREPES framework using Representer Landmarks and Nyström approximation.
result Direct audit of SSL latent space without supervision.
SOS-VAE improves generative models for scientific applications by correcting decoder bias.
problem Bias in generative parameters due to supervised learning in VAEs.
method Develops SOS-VAE framework to influence decoder for predictive latent representation.
result Ensures reliable generative parameters for scientific applications.
ExpBERT uses natural language explanations to improve text interpretation.
problem Improving text interpretation for relation extraction tasks.
method Fine-tuning BERT on MultiNLI to interpret natural language explanations.
result ExpBERT matches a BERT baseline but requires less labeled data and improves performance.
Proposes an adversarial algorithm to learn unbiased representations via HGR coefficient.
problem Learning fair representations without sensitive attribute information.
method Adversarial algorithm using Hirschfeld-Gebelein-Renyi (HGR) maximal correlation coefficient.
result Significant improvements in bias mitigation compared to existing methods.
Geometric framework explains and controls implicit bias in machine learning.
problem Understanding and controlling the selection of solutions in overparameterized models.
method Developed a theoretical and constructive framework based on geometric corrections induced by gradient noise and continuous symmetries of the loss.
result Computed the induced bias across various architectures and enabled inverse design to shape the bias.
UBM transfers bias mitigation from upstream to downstream tasks efficiently.
problem Bias in fine-tuned language models across various tasks.
method Apply bias mitigation to an upstream model, then fine-tune a downstream model on this mitigated model.
result UBM effects transfer to new downstream tasks, creating less biased models.
Chemical representations derived from deep learning are emerging as a powerful tool in areas such as drug discovery and materials innovation. Currently, this methodology has three major limitations - the cost of representation generation, risk of inherited bias, and the requirement for large amounts of data. We propose…
Efficiently scales continuous kernels with sparse Fourier domain learning.
problem High computational and memory demands, spectral bias in continuous kernels.
method Sparse learning in the Fourier domain.
result Efficient scaling of continuous kernels, reduced computational and memory requirements, mitigated spectral bias.
Most existing works on disentangled representation learning are solely built upon an marginal independence assumption: all factors in disentangled representations should be statistically independent. This assumption is necessary but definitely not sufficient for the disentangled representations without additional induc…
While current deep learning systems excel at tasks such as object classification, language processing, and gameplay, few can construct or modify a complex system such as a tower of blocks. We hypothesize that what these systems lack is a "relational inductive bias": a capacity for reasoning about inter-object relations…
MASC balances dataset representation using affinity clustering and distribution discrepancies.
problem Representation bias in datasets due to group imbalance.
method MASC uses affinity clustering and pairwise distribution discrepancies to balance non-protected and protected groups.
result MASC effectively debiases target datasets, comparable to existing methods.
The power of machine learning systems not only promises great technical progress, but risks societal harm. As a recent example, researchers have shown that popular word embedding algorithms exhibit stereotypical biases, such as gender bias. The widespread use of these algorithms in machine learning systems, from automa…
Softmax temperature influences model representation rank and performance.
problem Understanding and optimizing softmax function's impact on model representations.
method Investigated softmax function's role in deep neural networks, introduced rank deficit bias.
result Softmax temperature affects model representation rank and can improve performance.
Improves CNN robustness by reducing texture bias.
problem CNNs' reliance on local texture over global shape.
method Inspired by human vision, InfoDrop decorrelates model output from local texture.
result Enhanced robustness across various scenarios.
Paper debiases multiple word embedding biases simultaneously.
problem Reduction of multiple biases in word embeddings.
method Joint multiclass debiasing approach using Word Embeddings Association Test (WEAT).
result Demonstrates reduction or complete elimination of bias in word embeddings.
New model reduces matrix factorization bias, yielding truly low-rank solutions.
problem Gradient descent's implicit bias in matrix factorization.
method Introducing a new factorization model with constrained factors and diagonal components.
result The new model consistently exhibits a strong implicit bias, yielding truly low-rank solutions.
Paper finds sparse representation of functions using inverse scale space flow.
problem Finding sparse representation of L2 functions. method Inverse scale space flow to minimize L2 loss. result Convergence to optimal solution in ideal and noisy cases.
The paper connects neural collapse and low-rank bias in networks with L2 regularization.
problem Understanding the emergence of low-rank bias and neural collapse in L2-regularized networks.
method Unified theoretical framework linking TCV and rank of weight matrices, proving global optimality of DNC1, and establishing a benign landscape property.
result Zero TCV across intermediate layers minimizes representation cost under natural architectural constraints, and DNC1 is globally optimal.
This paper corrects climate model biases using a factor model approach.
problem Systematic biases in GCM outputs due to unobserved confounders.
method Factor model approach to learn latent confounders from historical data and apply them to enhance bias correction.
result Significant improvements in the accuracy of precipitation outputs.
New technique prevents Q-learning collapse by maximizing diversity among ensembles.
problem Value function collapse in ensemble Q-learning.
method Maximizing representation diversity through regularization.
result Regularized approach significantly outperforms existing methods.
New algorithm identifies causal relationships from graphs, even with selection bias.
problem Identifying causal relationships from graphs with selection bias.
method Developed a measure-theoretic version of Pearl's causal calculus and a sound, complete identification algorithm.
result General measure-theoretic version of causal calculus allows for identification of causal relationships under selection bias.
A method removes treatment-covariate dependence for counterfactual prediction without adversarial training.
problem Counterfactual prediction under assignment bias.
method Information-theoretic approach learning a stochastic representation Z to minimize mutual information with outcomes.
result The method performs favorably in likelihood, counterfactual error, and policy evaluation compared to adversarial baselines.
In real world systems, the predictions of deployed Machine Learned models affect the training data available to build subsequent models. This introduces a bias in the training data that needs to be addressed. Existing solutions to this problem attempt to resolve the problem by either casting this in the reinforcement l…