paper-with-me

홈 › Papers

Understanding the Failure Modes of Out-of-Distribution Generalization

2020-10-29 · ICLR 2021 1 · Vaishnavh Nagarajan, Anders Andreassen, Behnam Neyshabur

Empirical studies suggest that machine learning models often rely on features, such as the background, that may be spuriously correlated with the label only during training time, resulting in poor accuracy during test-time. In this work, we identify the fundamental factors that give rise to this behavior, by explaining why models fail this way {\em even} in easy-to-learn tasks where one would expect these models to succeed. In particular, through a theoretical study of gradient-descent-trained linear classifiers on some easy-to-learn tasks, we uncover two complementary failure modes. These modes arise from how spurious correlations induce two kinds of skews in the data: one geometric in nature, and another, statistical in nature. Finally, we construct natural modifications of image classification datasets to understand when these failure modes can arise in practice. We also design experiments to isolate the two failure modes when training modern neural networks on these datasets.

📄 PDF Abstract BibTeX arXiv:2010.15775

Code (1)

google-research/OOD-failures 공식 구현

Tasks

image-classificationImage ClassificationOut-of-Distribution Generalization

Similar Papers 제목 키워드 기반

Understanding Generalization through Decision Pattern Shift

2026-05-13 · Huiqi Deng, Yibo Li, Quanshi Zhang, Peng Zhang 외 arxiv

Understanding why deep neural networks (DNNs) fail to generalize to unseen samples remains a long-standing challenge. Existing studies mainly examine changes in externally observable factors such as data, representations…

Understanding the Failure Modes of Transformers through the Lens of Graph Neural Networks

2025-12-09 · Hunjae Lee arxiv

Transformers and more specifically decoder-only transformers dominate modern LLM architectures. While they have shown to work exceptionally well, they are not without issues, resulting in surprising failure modes and pre…

Graph Neural Network

Degradation Modeling and Prognostic Analysis Under Unknown Failure Modes

2024-02-29 · Ying Fu, Ye Kwon Huh, Kaibo Liu

Operating units often experience various failure modes in complex systems, leading to distinct degradation paths. Relying on a prognostic model trained on a single failure mode may lead to poor generalization performance…

Dimensionality Reduction

Discovering Failure Modes in Vision-Language Models using RL

2026-04-06 · Kanishk Jain, Qian Yang, Shravan Nayak, Parisa Kordjamshidi 외 arxiv

Vision-language Models (VLMs), despite achieving strong performance on multimodal benchmarks, often misinterpret straightforward visual concepts that humans identify effortlessly, such as counting, spatial reasoning, and…

Reinforcement LearningSpatial Reasoning

Aggregation Hides Out-of-Distribution Generalization Failures from Spurious Correlations

2025-10-28 · Olawale Salaudeen, Haoran Zhang, Kumail Alhamoud, Sara Beery 외 arxiv

Benchmarks for out-of-distribution (OOD) generalization frequently show a strong positive correlation between in-distribution (ID) and OOD accuracy across models, termed "accuracy-on-the-line." This pattern is often take…