No Subclass Left Behind: Fine-Grained Robustness in Coarse-Grained Classification Problems

By Nimit Sharad Sohoni, Jared A. Dunnmon, Geoffrey Angus, Albert Gu, Christopher R'e. Neural Information Processing Systems, 2020

306 citations recorded by the source corpus.

In real-world classification tasks, each class often comprises multiple finer-grained "subclasses." As the subclass labels are frequently unavailable, models trained using only the coarser-grained class labels often exhibit highly variable performance across different subclasses. This phenomenon, known as hidden stratification, has important consequences for models deployed in safety-critical applications such as medicine. We propose GEORGE, a method to both measure and mitigate hidden stratification even when subclass labels are unknown. We first observe that unlabeled subclasses are often separable in the feature space of deep models, and exploit this fact to estimate subclass labels for the training data via clustering techniques. We then use these approximate subclass labels as a form of noisy supervision in a distributionally robust optimization objective. We theoretically characterize the performance of GEORGE in terms of the worst-case generalization error across any subclass. We empirically validate GEORGE on a mix of real-world and benchmark image classification datasets, and show that our approach boosts worst-case subclass accuracy by up to 22 percentage points compared to standard training techniques, without requiring any information about the subclasses.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/8c96b865bbe1f597cf2c644e20ae46eab8e7caad

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

WILDS: A Benchmark of in-the-Wild Distribution Shifts (2020, unknown): “…al., 2019); other DRO algorithms that do not make use of explicit domain information, instead relying on unsupervised clustering (Oren et al., 2019; Sohoni et al., 2020) or worst-case subpopulations across all domains (Duchi et al., 2019); adaptive Lipschitz regularization (Cao et al., 2020);…”
Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision (2023, background): “However, it would also be interesting to directly study how well weak RMs can align policies; see e.g. Guo et al. (2024).”
Environment Inference for Invariant Learning (2020, methodology): “• GEORGE (Sohoni et al., 2020): This two-stage method seeks to learn the “hidden subclasses” by ﬁtting a latent cluster model to the (per-class) distribution of logits of a reference model.”
Distilling Model Failures as Directions in Latent Space (2022, unknown): “Sohoni et al. (2020) estimates subclass structure by clustering in the model’s latent space, while Liu et al. (2021) directly upweights examples that the model misclassiﬁed early in training.”
Data-centric Artificial Intelligence: A Survey (2023, unknown): “Moreover, to address hidden stratification, which occurs when each labeled class contains multiple semantically distinct subclasses, GEORGE [208] employs clustering algorithms to slide data across different subclasses.”