arXiv:2608. 16947v1 Announce Type: cross Abstract: Huang, Lou, and Xiao introduced Dynamic Mixture-of-Experts Serving and gave an O(sqrt(log k))-competitive randomized algorithm for its integral primal problem, where k is the number of replica GPUs beyond the mandatory copy of each expert.
arXiv:2608.16947v2 Announce Type: replace-cross Abstract: Dynamic Mixture-of-Experts Serving allocates k replica GPUs among m experts as workloads change. At each round, the online algorithm sees the...
arXiv:2608. 12134v1 Announce Type: cross Abstract: We study nonnegative submodular maximization subject to a general matroid when the offline algorithm is given an arbitrary controlled value oracle.
arXiv:2603. 10346v2 Announce Type: replace Abstract: We study the fixed-budget best-arm identification (BAI) problem in non-stationary linear bandits.
arXiv:2608. 01616v1 Announce Type: new Abstract: Competitive analysis is central to the study of online algorithms, but upper bounds are often highly problem-specific.
The paper studies online convex optimization when the learner can only query an exact linear optimization oracle. It establishes a dimension‑free minimax expected regret bound of θ(GD max{√T, T/(1+min{Q,BT})^{1/4}}) for convex G‑Lipschitz losses, where Q is the total oracle budget and B the per‑round limit. The authors provide matching lower and upper bounds, showing how strict per‑round or total‑budget constraints affect the achievable regret, and extend the analysis to smooth losses with curvature‑dependent bounds.
The paper revisits the Successive Rejects (SR) algorithm for best‑arm identification, showing that although SR appears non‑adaptive due to fixed phase lengths, it achieves an optimal worst‑instance error exponent for Gaussian arms with common variance. The authors formulate an optimization problem to design a phase schedule that maximizes the normalized error exponent, leading to a new variant called H1‑balanced Successive Rejects (H1‑SR). H1‑SR attains at least 11/12 of the optimal normalized rate for all K ≥ 2, approaching the optimum as K grows, and experiments demonstrate significant performance gains over the original SR while highlighting trade‑offs in early elimination strategies.
arXiv:2609.10529v1 Announce Type: cross Abstract: We prove the gap-entropy conjecture for fixed-confidence best-arm identification with independent unit-variance Gaussian arms, means in $[0,1]$, and...
arXiv:2608.12134v2 Announce Type: replace-cross Abstract: We study nonnegative submodular maximization on $n$ elements subject to a general matroid of rank $k$, when the offline algorithm is given an...
We prove the gap-entropy conjecture for fixed-confidence best-arm identification with independent unit-variance Gaussian arms, means in $[0,1]$, and a unique optimal arm. For each suboptimal arm $i$,...
The paper introduces a new primal–dual algorithm for episodic adversarial linear constrained Markov decision processes (CMDPs) with unknown transitions. It achieves a rate‑optimal ×O(√K) regret and cumulative constraint violation, improving upon the previous ×O(K^{3/4}) bound and eliminating the need for Slater’s condition. The method combines adaptive FTRL, contracted value estimation, and an exponential Lyapunov function, enabling uniform concentration over the value function class and computational efficiency independent of the state‑space size.
arXiv:2610.01951v1 Announce Type: cross Abstract: Top-two algorithms are simple and effective for fixed-confidence best-arm identification, but their sharp non-asymptotic behavior is still not well u...
The paper introduces Dec-BFTRL, a decentralized algorithm for online optimization of upper-linearizable payoffs with efficient separation access, targeting continuous diminishing-return submodular maximization. Each agent evaluates its action against the average of local objectives, projects via an approximate gauge, exchanges a cumulative surrogate-gradient dual state, and uses a local HybridNewton step to minimize its BFTRL potential. The method achieves an expected network-aggregate regret of “~O(√T)” while requiring T neighbor-mixing steps and ~O(T) separation-oracle calls per agent, and provides four wrapper instantiations for three DR-submodular problems.