Unverified50% confidenceFactExact time
This upper bound exhibits a 1/ε rate when ε < β, and its dependence on H is the same as in unregularized RL (H^3); when ε is large, the sample complexity is inversely proportional to β.
1
Sources
50%
Confidence
Long-term
Relevance
11/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedFor Gaussian arms with a common variance, H₁-SR achieves at least 11/12 of the optimal normalized rate among all adaptive algorithms, and this ratio holds for any K≥2.71% similarUnverifiedThe momentum method introduces a velocity variable v that accumulates historical gradients: v = βv + (1-β)∇θL, with the parameter update θ = θ - αv, where β is typically set to 0.9.68% similarUnverifiedWhen α > ν, the lower bound on the minimum number of simulation evaluations required by any algorithm to simultaneously identify the near-optimal regions for all tolerances is Ω(η^{-(2+d/ν)} min{1, KT^{d/α}}).66% similarUnverifiedThe new sample complexity bound matches the known lower bound in its exponential dependence on |β|/(1-γ).65% similarUnverifiedThere exists a universal constant ε0 > 0 such that for any finite residual frame, T_3 ≤ (5/6 - ε0)E holds.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1158401API
curl https://kongchang.com/api/v1/knowledge/claims/1158401MCP
get_claim(id=1158401)