0:00 / 0:00
#discrepancytheory#informationtheory#komlosconjecture
Probability, Information, and Discrepancy
Simons Institute for the Theory of Computing
77K subscribers • 30 videos on Hallway • University channel
- One of the most-watched uploads on Simons Institute, a conference or university channel Hallway lists. Nobody has recommended it in the communities yet, so it ranks below community picks.
- Presented by Roman Vershynin, UC Irvine
Advanced. Deep dive for people who know the field.
5.2K views2 weeks ago
Computer scienceAI and machine learningComputation and complexityMath for programmersNeural networks and deep learning
Roman Vershynin discusses improving random samples through discrepancy theory and how these ideas relate to information theory and AI.
Roman Vershynin (UC Irvine) simons.berkeley.edu/talks/roman-vershynin-uc-irvine-2026-10-… Probabilistic Phenomena on Graphs, Groups, and Manifolds Can we improve a random sample by modifying just a tiny fraction of its points? This fascinating question brings together statistics (which motivates it), probability theory (which gives us the tools), and discrepancy theory (where it takes a deterministic form). I’ll describe a series of joint works with Gleb Smirnov that led us on an interesting detour into information theory. Our central idea was recently used by AI to solve the long-standing Komlós conjecture.Why it's here
A top upload from a listed channelSimons Institute is on Hallway's list of conference and university channels, and this upload is one of its most-watched talks or lectures. Nobody has recommended it in the communities yet, so it ranks below community picks.