A Closed-Form Upper Bound for Admissible Learning-Rate Steps in Belief-Space Dynamics
Not on the map. Why?
Abstract (arXiv)
Learning-rate steps are usually treated as hyperparameters. This paper isolates a local beliefspace calculation: when an update is modeled as a projected forward step on the probability simplex, admissibility means contractivity in the natural KL/Bregman geometry. Under this model, the upper bound of an admissible step is not a tuning slogan but a formula.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 7 of its 12 references that are in the map. First posted on arXiv 7 May 2026.
Builds on
7 of 12 references.
References · 7
- 2023Optimal Linear Decay Learning Rate Schedules and Further Refinementscited by 43
- 2020Adaptive Inertia: Disentangling the Effects of Adaptive Learning Rate and Momentumcited by 40
- 2015Layer-Specific Adaptive Learning Rates for Deep Networkscited by 19
- 2018Beyond Convexity -- Contraction and Global Convergence of Gradient Descentcited by 17
- 2022Conformal Mirror Descent with Logarithmic Divergencescited by 3
- 2021A note on the optimal convergence rate of descent methods with fixed step sizes for smooth strongly convex functionscited by 2
- 2023Convergence Rate Bounds for the Mirror Descent Method: IQCs, Popov Criterion and Bregman Divergencecited by 1
Built on by
Not cited yet.