On the Implicit Geometry of Cross-Entropy Parameterizations for Label-Imbalanced Data

Tina Behnia, Ganesh Ramachandra Kini, Vala Vakilian, Christos Thrampoulidis

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations


Various logit-adjusted parameterizations of the cross-entropy (CE) loss have been proposed as alternatives to weighted CE for training large models on label-imbalanced data far beyond the zero train error regime. The driving force behind those designs has been the theory of implicit bias, which for linear(ized) models, explains why they successfully induce bias on the optimization path towards solutions that favor minorities. Aiming to extend this theory to non-linear models, we investigate the implicit geometry of classifiers and embeddings that are learned by different CE parameterizations. Our main result characterizes the global minimizers of a non-convex cost-sensitive SVM classifier for the unconstrained features model, which serves as an abstraction of deep-nets. We derive closed-form formulas for the angles and norms of classifiers and embeddings as a function of the number of classes, the imbalance and the minority ratios, and the loss hyperparameters. Using these, we show that logit-adjusted parameterizations can be appropriately tuned to learn symmetric geometries irrespective of the imbalance ratio. We complement our analysis with experiments and an empirical study of convergence accuracy in deep-nets.
Original languageEnglish (US)
Title of host publication26th International Conference on Artificial Intelligence and Statistics, AISTATS 2023
PublisherML Research Press
Number of pages24
StatePublished - Jan 1 2023
Externally publishedYes


Dive into the research topics of 'On the Implicit Geometry of Cross-Entropy Parameterizations for Label-Imbalanced Data'. Together they form a unique fingerprint.

Cite this