mlsanityData to 14 Oct 2026

When Data Imbalance Helps: Robust Generalization Through Shortcut Saturation

arXiv 2607.10116 ↗First posted 11 Jul 2026cs.LG cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

We study robust generalization under spurious correlations: tasks where a shortcut feature is correlated with the true label in training but anti-correlated in an adversarial held-out split. Varying the spurious ratio r (the fraction of training examples where shortcut = true label) and model capacity, we find a counterintuitive result: data imbalance promotes generalization in sufficiently capable models. On a synthetic task where the true label is sum parity of an integer sequence and the shortcut is the parity of the maximum-valued element, a 2-layer, 2-head transformer generalized (reached 100% adversarial accuracy) in 0% of seeds at r=0.50 but 77% of seeds at r=0.90. The effect is absent in 1-layer models, where imbalance instead traps the model on the shortcut. Through mechanistic analysis -- gradient conflict dynamics, circuit evolution, and QK/OV circuit ablations -- we characterize a mechanistic pathway consistent with imbalance promoting generalization.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 14 of its 14 references that are in the map. First posted on arXiv 11 Jul 2026.

Builds on

14 of 14 references.

References · 14

  1. 2020Shortcut Learning in Deep Neural Networkscited by 1,969
  2. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,247
  3. 2022In-context Learning and Induction Headscited by 964
  4. 2023Progress measures for grokking via mechanistic interpretabilitycited by 839
  5. 2022Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasetscited by 629
  6. 2020The Pitfalls of Simplicity Bias in Neural Networkscited by 351
  7. 2020An Investigation of Why Overparameterization Exacerbates Spurious Correlationscited by 310
  8. 2022Hidden Progress in Deep Learning: SGD Learns Parities Near the Computational Limitcited by 187
  9. 2023The Clock and the Pizza: Two Stories in Mechanistic Explanation of Neural Networkscited by 156
  10. 2023On the Foundations of Shortcut Learningcited by 65
  11. 2021Domain Generalization via Gradient Surgerycited by 60
  12. 2020Class-Wise Difficulty-Balanced Loss for Solving Class-Imbalancecited by 23
2 more
  1. 2025Uncovering Memorization Effect in the Presence of Spurious Correlationscited by 20
  2. 2025Do We Always Need the Simplicity Bias? Looking for Optimal Inductive Biases in the Wildcited by 13

Built on by

Not cited yet.