Lower Bounds for Linear-Oracle Online Learning

arXiv:2609.38375 · stat.ML, cs.LG, math.OC · Submitted 2026-09-29 · Read on arXiv

stat.ML, cs.LG, math.OC

Submitted: 2026-09-29

Updated: 2026-09-29

Terminology

• Regret — 32× in this paper · explained in 3 episodes
• Ambient Dimension — 3× in this paper · explained in 1 episode
• Verifier — 2× in this paper · explained in 2 episodes
• Embeddings — 1× in this paper · explained in 2 episodes
• Semidefinite Program (SDP) — 1× in this paper · explained in 2 episodes

Sources

• Lower Bounds for Linear Minimization Oracle Methods Optimizing over Strongly Convex Sets
• Lower Bounds for Frank-Wolfe on Strongly Convex Sets
• Curvature-Dependent Lower Bounds for Frank-Wolfe
• The Complexity of Large-scale Convex Programming under a Linear Optimization Oracle
• Exploiting the Curvature of Feasible Sets for Faster Projection-Free Online Learning

Related papers

• Behavior of prediction performance metrics with rare events
• Optimal Estimation of Generic Dynamics by Path-Dependent Neural Jump ODEs
• A Posterior-Dynamics Framework for Imaging Inverse Problems with Pretrained Diffusion Priors
• Efficient quadratic entropy with distance sketches
• Embedding-Bias in Conditional Independence Testing
• From Log-Odds to Shapley Values: An Explanatory Geometry for the Weighted Naive Bayes Classifier