The Group Robustness is in the Details: Revisiting Finetuning under Spurious Correlations

• URL: http://arxiv.org/abs/2407.13957v1
• Date: Fri, 19 Jul 2024 00:34:03 GMT
• Title: The Group Robustness is in the Details: Revisiting Finetuning under Spurious Correlations
• Authors: Tyler LaBonte, John C. Hill, Xinchen Zhang, Vidya Muthukumar, Abhishek Kumar,
• Abstract summary: In this paper, we identify surprising and nuanced behavior of finetuned models on worst-group accuracy. We first show that the commonly used class-balancing techniques of mini-batch upsampling and loss upweighting can induce a decrease in worst-group accuracy. Next, we show that scaling pretrained models is generally beneficial for worst-group accuracy, but only in conjuction with appropriate class-balancing.
• Score: 8.844894807922902
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Modern machine learning models are prone to over-reliance on spurious correlations, which can often lead to poor performance on minority groups. In this paper, we identify surprising and nuanced behavior of finetuned models on worst-group accuracy via comprehensive experiments on four well-established benchmarks across vision and language tasks. We first show that the commonly used class-balancing techniques of mini-batch upsampling and loss upweighting can induce a decrease in worst-group accuracy (WGA) with training epochs, leading to performance no better than without class-balancing. While in some scenarios, removing data to create a class-balanced subset is more effective, we show this depends on group structure and propose a mixture method which can outperform both techniques. Next, we show that scaling pretrained models is generally beneficial for worst-group accuracy, but only in conjuction with appropriate class-balancing. Finally, we identify spectral imbalance in finetuning features as a potential source of group disparities -- minority group covariance matrices incur a larger spectral norm than majority groups once conditioned on the classes. Our results show more nuanced interactions of modern finetuned models with group robustness than was previously known. Our code is available at https://github.com/tmlabonte/revisiting-finetuning.

Related papers

• Multitask Learning Can Improve Worst-Group Outcomes [76.92646345152788]
  Multitask learning (MTL) is one such widely used technique. We propose to modify standard MTL by regularizing the joint multitask representation space. We find that our regularized MTL approach emphconsistently outperforms JTT on both average and worst-group outcomes.
  arXiv  Detail & Related papers  (2023-12-05T21:38:24Z)
• Bias Amplification Enhances Minority Group Performance [10.380812738348899]
  We propose BAM, a novel two-stage training algorithm. In the first stage, the model is trained using a bias amplification scheme via introducing a learnable auxiliary variable for each training sample. In the second stage, we upweight the samples that the bias-amplified model misclassifies, and then continue training the same model on the reweighted dataset.
  arXiv  Detail & Related papers  (2023-09-13T04:40:08Z)
• Ranking & Reweighting Improves Group Distributional Robustness [14.021069321266516]
  We propose a ranking-based training method called Discounted Rank Upweighting (DRU) to learn models that exhibit strong OOD performance on the test data. Results on several synthetic and real-world datasets highlight the superior ability of our group-ranking-based (akin to soft-minimax) approach in selecting and learning models that are robust to group distributional shifts.
  arXiv  Detail & Related papers  (2023-05-09T20:37:16Z)
• Simplicity Bias Leads to Amplified Performance Disparities [8.60453031364566]
  We show that SGD-trained models have a bias towards simplicity, leading them to prioritize learning a majority class. A model may prioritize any class or group of the dataset that it finds simple-at the expense of what it finds complex.
  arXiv  Detail & Related papers  (2022-12-13T15:24:41Z)
• Outlier-Robust Group Inference via Gradient Space Clustering [50.87474101594732]
  Existing methods can improve the worst-group performance, but they require group annotations, which are often expensive and sometimes infeasible to obtain. We address the problem of learning group annotations in the presence of outliers by clustering the data in the space of gradients of the model parameters. We show that data in the gradient space has a simpler structure while preserving information about minority groups and outliers, making it suitable for standard clustering methods like DBSCAN.
  arXiv  Detail & Related papers  (2022-10-13T06:04:43Z)
• Take One Gram of Neural Features, Get Enhanced Group Robustness [23.541213868620837]
  Predictive performance of machine learning models trained with empirical risk minimization can degrade considerably under distribution shifts. We propose to partition the training dataset into groups based on Gram matrices of features extracted by an identification'' model. Our approach not only improves group robustness over ERM but also outperforms all recent baselines.
  arXiv  Detail & Related papers  (2022-08-26T12:34:55Z)
• Correct-N-Contrast: A Contrastive Approach for Improving Robustness to Spurious Correlations [59.24031936150582]
  Spurious correlations pose a major challenge for robust machine learning. Models trained with empirical risk minimization (ERM) may learn to rely on correlations between class labels and spurious attributes. We propose Correct-N-Contrast (CNC), a contrastive approach to directly learn representations robust to spurious correlations.
  arXiv  Detail & Related papers  (2022-03-03T05:03:28Z)
• Just Train Twice: Improving Group Robustness without Training Group Information [101.84574184298006]
  Standard training via empirical risk minimization can produce models that achieve high accuracy on average but low accuracy on certain groups. Prior approaches that achieve high worst-group accuracy, like group distributionally robust optimization (group DRO) require expensive group annotations for each training point. We propose a simple two-stage approach, JTT, that first trains a standard ERM model for several epochs, and then trains a second model that upweights the training examples that the first model misclassified.
  arXiv  Detail & Related papers  (2021-07-19T17:52:32Z)
• Examining and Combating Spurious Features under Distribution Shift [94.31956965507085]
  We define and analyze robust and spurious representations using the information-theoretic concept of minimal sufficient statistics. We prove that even when there is only bias of the input distribution, models can still pick up spurious features from their training data. Inspired by our analysis, we demonstrate that group DRO can fail when groups do not directly account for various spurious correlations.
  arXiv  Detail & Related papers  (2021-06-14T05:39:09Z)
• Model Patching: Closing the Subgroup Performance Gap with Data Augmentation [50.35010342284508]
  We introduce model patching, a framework for improving robustness of machine learning models. Model patching encourages the model to be invariant to subgroup differences, and focus on class information shared by subgroups. We instantiate model patching with CAMEL, which (1) uses a CycleGAN to learn the intra-class, inter-subgroup augmentations, and (2) balances subgroup performance using a theoretically-motivated consistency regularizer. We demonstrate CAMEL's effectiveness on 3 benchmark datasets, with reductions in robust error up to 33% relative to the best baseline.
  arXiv  Detail & Related papers  (2020-08-15T20:01:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.