Block-Norm Geometries for Online Mirror Descent with Sparse Losses

arXiv:2602.13177 · math.OC, cs.DS, cs.LG · Submitted 2026-02-13 · Read on arXiv

math.OC, cs.DS, cs.LG

Submitted: 2026-02-13

Updated: 2026-09-11

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

Regret — 150× in this paper · explained in 3 episodes
Loss Function — 62× in this paper · explained in 2 episodes
Sparsity — 21× in this paper · explained in 4 episodes
Online Convex Optimization — 19× in this paper
Online Mirror Descent — 15× in this paper
Multiplicative Weight — 6× in this paper
Online Exponentiated Gradient — 6× in this paper
Online Projected Gradient Descent — 6× in this paper
Adaptation — 1× in this paper · explained in 1 episode
Adaptive Algorithm — 1× in this paper · explained in 1 episode
Adaptivity — 1× in this paper · explained in 1 episode
Aggregation — 1× in this paper · explained in 1 episode
Benchmark — 1× in this paper · explained in 2 episodes
Core — 1× in this paper · explained in 2 episodes
Generalization — 1× in this paper · explained in 13 episodes
Minimax Regret — 1× in this paper · explained in 2 episodes

Sources

Potential-Function Proofs for First-Order Methods

Related papers