Paper: minimum enclosing Bregman balls solvable as linear-programming-type problems
FrnkNlsn · x · 2026-09-15
The paper 'On the smallest enclosing information disk' shows that the minimum enclosing Bregman ball problem — covering Euclidean distance and KL divergence as special cases — can be solved as a linear-programming-type problem, offering a unified computational route for Bregman-divergence-based tasks via standard LP solvers.
More from Research
- Lecture slides online: Reinforcement Learning for Autonomous Driving at ELLIS summer school — abursuc · 2026-09-15
- New arXiv paper invites mathematicians to tackle AI safety, field by field — stevenstrogatz · 2026-09-15
- Orthrus Study: Lossless Speculative Decoding Holds Only at High Numerical Precision — Ilya Koziev · 2026-09-15
- MIT team's Recursive Meta-Intelligence: AI builds its own instruments to crack metamaterials — ProfBuehlerMIT · 2026-09-15
- Ground Truth Is Rarely Ground Truth in AI for Science, Researcher Warns — bravo_abad · 2026-09-15
- Pitch: train a generative game model on RL agents playing millions of times, ship the weights — theteknosaur · 2026-09-15