JTT improves model worst-group accuracy without group annotations.
problem Low worst-group accuracy in standard ERM models with spurious correlations.
method Two-stage approach: first ERM, then upweight misclassified examples.
result JTT closes 75% of the gap in worst-group accuracy compared to group DRO.
Investigates upsampling vs. upweighting for balanced training on skewed datasets.
problem Balancing training on heavily imbalanced datasets with scarce data.
method Theoretical and empirical analysis of upsampling and upweighting strategies.
result Upsampling and upweighting diverge under stochastic gradient descent, with upsampling leading to faster convergence but higher overfitting risk.
Our method upweights easy samples to mitigate forgetting in fine-tuning.
problem Catastrophic forgetting in fine-tuning pre-trained models.
method Sample weighting based on pre-trained model's losses.
result Our method reduces forgetting by up to 0.8% on MetaMathQA while preserving more accuracy on pre-training datasets.
Data augmentation methods improve worst-case model performance.
problem Ensuring fair predictions across subpopulations in large models.
method Linear last layer retraining with data augmentation techniques.
result Optimal worst-group accuracy achieved for Gaussian latent representation distribution.
Discrepancy between training and testing domains is a fundamental problem in the generalization of machine learning techniques. Recently, several approaches have been proposed to learn domain invariant feature representations through adversarial deep learning. However, label shift, where the percentage of data in each …
RAD improves robustness to domain annotation noise without explicit domain annotations.
problem Robustness to domain annotation noise in training data.
method Regularized Annotation of Domains (RAD) for last layer retraining.
result RAD outperforms state-of-the-art methods even with 5% noise in training data.
Unweighted matrix factorization can match or outperform weighted methods in recommender systems.
problem Improving recommendation performance with matrix factorization on implicit feedback data.
method Systematic study of various weighting schemes and matrix factorization algorithms.
result Training with unweighted data can perform comparably to, and sometimes outperform, training with weighted data.
Overparameterized models can worsen minority group errors even when overall test error improves.
problem Overparameterization exacerbates spurious correlations, harming minority groups.
method Simulations and experiments on image datasets, theoretical analysis of linear models.
result Subsampling the majority group can achieve low minority error in overparameterized models.
A new method for graph-structured data improves transformer performance by incorporating topology.
problem Improving transformer performance on graph-structured data.
method Parameterizing topological masks as a learnable function of a weighted adjacency matrix, approximated with graph random features.
result Efficient masking algorithms provide strong performance gains for tasks on image and point cloud data.
Proposes learning regularization strength directly from data.
problem Computational expense and data reduction in grid search for deep learning hyperparameters.
method Modified Evidence Lower Bound (ELBo) objective for model selection on full training set.
result Comparable heldout accuracy to grid search with less compute time.
A novel algorithm for unbiased graph kernel estimation with subquadratic time complexity.
problem Efficient estimation of graph kernels for large networks.
method Random walk-based algorithm with modulation function parameterized by neural network.
result Higher-quality kernel estimates and efficient scalable learning on larger networks.
MP-Boost boosts accuracy faster and more interpretable than AdaBoost.
problem Developing a faster, more interpretable boosting method.
method Adaptive selection of minipatches (small subsets of instances and features) at each iteration.
result Achieves comparable accuracy to AdaBoost and gradient boosting but faster and more interpretable.
AFR simplifies reducing reliance on spurious features, improving model performance.
problem Reducing reliance on spurious features for out-of-distribution generalization.
method Automatic Feature Reweighting (AFR) updates the model with a weighted loss.
result AFR improves model performance on benchmarks with minimal compute.
CrossWalk enhances fairness in graph algorithms by biasing random walks.
problem Fairness in machine learning systems applied to graphs.
method Bias random walks to cross group boundaries by upweighting edges.
result Enhances fairness in various graph algorithms with minimal performance loss.
The paper proposes a method to learn hyperparameters without validation sets, improving efficiency and accuracy.
problem Training large models on limited data to avoid overfitting and reduce validation set usage.
method Gradient-based learning of hyperparameters via a data-emphasized evidence lower bound (ELBO) objective.
result The data-emphasized ELBO reduces hyperparameter search time from 88+ hours to under 3 hours while maintaining comparable accuracy.
Logit correction improves model performance by correcting spurious correlations.
problem Spurious correlations lead to poor model performance during inference.
method Proposes logit correction (LC) loss to mitigate spurious correlations.
result LC loss outperforms state-of-the-art solutions by 5.5% absolute improvement.
A self-supervised debiasing method using rank regularization mitigates spurious correlations in neural networks.
problem Spurious correlations cause biases in deep neural networks, affecting generalization.
method Spectral analysis of latent representations, rank regularization, self-supervised pretraining, debiasing of downstream tasks.
result The proposed framework significantly improves generalization performance and outperforms supervised debiasing approaches.
The paper analyzes how forgetting in LLMs is linked to simple task-upstream example associations.
problem Forgetting of upstream knowledge in fine-tuned LLMs.
method Empirical analysis of forgotten examples in N upstream examples after M new tasks, using low-rank matrix approximation. result Forgetting can be predicted efficiently using matrix completion over empirical associations.
The paper tackles distribution-free prediction intervals for multi-source data.
problem Challenges in achieving valid inferences due to distribution shifts and privacy concerns.
method Derives efficient influence functions, incorporates machine learning, and proposes data-adaptive strategies.
result Achieves parametric rates of convergence to nominal coverage probabilities for prediction intervals.
MARVEL curbs memorization of noisy labels in deep nets.
problem Noisy labels degrade deep net performance.
method MARVEL tracks classification margins to identify and abandon noisy instances.
result MARVEL outperforms baselines on noisy datasets.
This paper improves model robustness to underrepresented groups using ranking metrics and reweighting.
problem Underrepresented groups suffer from low accuracy in models trained via ERM.
method Proposes Discounted Cumulative Gain (DCG) and Discounted Rank Upweighting (DRU) methods.
result Models trained with DRU show superior generalization to unseen groups.
Boosts causal discovery by dynamically reweighting samples to learn better DAGs.
problem Overfitting to easier-to-fit samples and violating homogeneity assumptions in causal discovery.
method Adaptive sample reweighting via ReScore function to upweight and downweight samples based on fitting quality.
result Consistent and significant boosts in structure learning performance on synthetic and real-world datasets.
Recommender systems widely use implicit feedback such as click data because of its general availability. Although the presence of clicks signals the users' preference to some extent, the lack of such clicks does not necessarily indicate a negative response from the users, as it is possible that the users were not expos…