BREEDS: Benchmarks for Subpopulation Shift

Abstract

We develop a methodology for assessing the robustness of models to subpopulation shift---specifically, their ability to generalize to novel data subpopulations that were not observed during training. Our approach leverages the class structure underlying existing datasets to control the data subpopulations that comprise the training and test distributions. This enables us to synthesize realistic distribution shifts whose sources can be precisely controlled and characterized, within existing large-scale datasets. Applying this methodology to the ImageNet dataset, we create a suite of subpopulation shift benchmarks of varying granularity. We then validate that the corresponding shifts are tractable by obtaining human baselines for them. Finally, we utilize these benchmarks to measure the sensitivity of standard model architectures as well as the effectiveness of off-the-shelf train-time robustness interventions. Code and data available at https://github.com/MadryLab/BREEDS-Benchmarks .

References in corpus (16)

• Distributed Representations of Words and Phrases and their Compositionality
• DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition
• ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness
• Domain Generalization via Invariant Feature Representation
• The Open Images Dataset V4: Unified image classification, object detection, and visual relationship detection at scale
• Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalization
• What makes ImageNet good for transfer learning?
• Invariant Risk Minimization
• LEAF: A Benchmark for Federated Settings
• Exploring the Landscape of Spatial Robustness
• Measuring Robustness to Natural Distribution Shifts in Image Classification
• Adversarial Examples Are a Natural Consequence of Test Error in Noise
• Benchmarking Neural Network Robustness to Common Corruptions and Surface Variations
• Does Distributionally Robust Supervised Learning Give Robust Classifiers?
• Testing Robustness Against Unforeseen Adversaries
• From ImageNet to Image Classification: Contextualizing Progress on Benchmarks

Cited by in corpus (7)

• On the Opportunities and Risks of Foundation Models
• Evaluating Model Robustness and Stability to Dataset Shift
• Model-Based Domain Generalization
• An Information-theoretic Approach to Distribution Shifts
• Clustering Effect of (Linearized) Adversarial Robust Models
• Data Augmentation via Structured Adversarial Perturbations
• Robustness Evaluation of Transformer-based Form Field Extractors via Form Attacks