The Price of Hidden Curvature: An $\widetildeΩ (d^{5/4} \sqrt{T})$ Lower Bound for Bandit Convex Optimization

• URL: http://arxiv.org/abs/2607.18652v2
• Date: Wed, 22 Jul 2026 18:39:47 GMT
• Title: The Price of Hidden Curvature: An $\widetildeΩ (d^{5/4} \sqrt{T})$ Lower Bound for Bandit Convex Optimization
• Abstract summary: We show that a $widetilde(d5/4sqrt T)$ lower bound on the minimax expected regret of bandit convex optimization of $1$-Lipschitz functions grows faster than $dsqrtT$.<n>This presents the first nontrivial regret lower bound that grows faster than $dsqrtT$ for this problem.
• Score: 4.774251262663345
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We establish a $\widetildeΩ(d^{5/4}\sqrt T)$ lower bound on the minimax expected regret of stochastic bandit convex optimization of $1$-Lipschitz functions on the Euclidean ball. This presents the first nontrivial regret lower bound that grows faster than $d\sqrt{T}$ for this problem, establishing that stochastic bandit convex optimization is fundamentally harder than linear bandits. The hard class of convex functions we construct takes the following form in dimension $2d$: for an action $a = (a^1,a^2) \in \mathbb{B}^{2d}_2$, each function is the scaled soft maximum of a "tube", $r^{-1} \| W^\star a^1 - \frac{r}{8\varepsilon} a^2 \|_2$ (hyperparameterized by $\varepsilon,r$), and a squared distance function, $\frac12 \| a^1 - u^\star \|_2^2 - \frac12 \| u^\star \|_2^2$. Here, $W^\star \in \mathbb{R}^{d \times d}$ is an unknown linear transformation, and $u^\star \in \mathbb{R}^{d}$ is an unknown vector which must be learned to minimize the function. Observations are informative about $u^\star$ only when the learner's action lies near the tube determined by $W^\star$, satisfying $a^2 \approx \frac{8\varepsilon}{r} W^\star a^1$: thus the learner must either find this tube without knowing $W^\star$, or spend observations learning useful directions of $W^\star$. Formally, our regret analysis exploits this tradeoff by bounding the posterior spread of Fisher information matrices obtained under an adaptive sequence of actions. Together, these ingredients give a sample complexity lower bound of $\widetildeΩ(d^{5/2}/\varepsilon^2)$ to find an $\varepsilon$-optimal action, which translates to an $\widetildeΩ (d^{5/4} \sqrt{T})$ regret lower bound. We also extend this lower bound to the unconstrained setting where the action space is $\mathbb{R}^d$.

Related papers

• On the Power of Adaptivity for $\varepsilon$-Best Arm Identification in Linear Bandits [8.86974002494041]
  We study the minimax sample complexity of $varepsilon$-best arm identification in linear bandits.<n>The goal is to output an arm $widehatxinmathcalX$ such that $langle ge max_xinmathcalX langle x,rangle ge max_xinmathcalX langle x,rangle ge max_xinmathcalX langle x
  arXiv  Detail & Related papers  (2026-05-15T06:35:31Z)
• From Average Sensitivity to Small-Loss Regret Bounds under Random-Order Model [26.860985092865203]
  We study online learning in the random-order model, where the multiset of loss functions is chosen adversarially but revealed in a uniformly random order.<n>Our approach sheds light on the power of sparsification and related techniques in establishing small-loss regret bounds in the random-order model.
  arXiv  Detail & Related papers  (2026-02-10T06:46:01Z)
• Near-Optimal Convergence of Accelerated Gradient Methods under Generalized and $(L_0, L_1)$-Smoothness [57.93371273485736]
  We study first-order methods for convex optimization problems with functions $f$ satisfying the recently proposed $ell$-smoothness condition $||nabla2f(x)|| le ellleft(||nabla f(x)||right),$ which generalizes the $L$-smoothness and $(L_0,L_1)$-smoothness.
  arXiv  Detail & Related papers  (2025-08-09T08:28:06Z)
• Improved Regret Bounds for Linear Bandits with Heavy-Tailed Rewards [38.53963338592248]
  We improve both upper and lower bounds on the minimax regret compared to prior work.<n>We propose a new elimination-based algorithm guided by experimental design.<n>We show that for some geometries, such as $l_p$-norm balls for $p le 1 + epsilon$, we can further reduce the dependence on $d$.
  arXiv  Detail & Related papers  (2025-06-05T09:07:26Z)
• Optimal Sketching for Residual Error Estimation for Matrix and Vector Norms [50.15964512954274]
  We study the problem of residual error estimation for matrix and vector norms using a linear sketch. We demonstrate that this gives a substantial advantage empirically, for roughly the same sketch size and accuracy as in previous work. We also show an $Omega(k2/pn1-2/p)$ lower bound for the sparse recovery problem, which is tight up to a $mathrmpoly(log n)$ factor.
  arXiv  Detail & Related papers  (2024-08-16T02:33:07Z)
• Efficient Continual Finite-Sum Minimization [52.5238287567572]
  We propose a key twist into the finite-sum minimization, dubbed as continual finite-sum minimization. Our approach significantly improves upon the $mathcalO(n/epsilon)$ FOs that $mathrmStochasticGradientDescent$ requires. We also prove that there is no natural first-order method with $mathcalOleft(n/epsilonalpharight)$ complexity gradient for $alpha 1/4$, establishing that the first-order complexity of our method is nearly tight.
  arXiv  Detail & Related papers  (2024-06-07T08:26:31Z)
• Tackling Heavy-Tailed Rewards in Reinforcement Learning with Function Approximation: Minimax Optimal and Instance-Dependent Regret Bounds [26.277745106128197]
  In this work, we address the challenge of such rewards in reinforcement learning with linear function approximation. We first design an algorithm, textscHeavy-OFUL, for heavy-tailed linear bandits, achieving an emphinstance-dependent $T$-round regret of $tildeObig. Our result is achieved via a novel self-normalized concentration inequality that may be of independent interest in handling heavytailed noise in general online regression problems.
  arXiv  Detail & Related papers  (2023-06-12T02:56:09Z)
• Low-degree learning and the metric entropy of polynomials [44.99833362998488]
  We prove that any (deterministic or randomized) algorithm which learns $mathscrF_nd$ with $L$-accuracy $varepsilon$ requires at least $Omega(sqrtvarepsilon)2dlog n leq log mathsfM(mathscrF_n,d,|cdot|_L,varepsilon) satisfies the two-sided estimate $$c (1-varepsilon)2dlog
  arXiv  Detail & Related papers  (2022-03-17T23:52:08Z)
• Optimal Spectral Recovery of a Planted Vector in a Subspace [80.02218763267992]
  We study efficient estimation and detection of a planted vector $v$ whose $ell_4$ norm differs from that of a Gaussian vector with the same $ell$ norm. We show that in the regime $n rho gg sqrtN$, any spectral method from a large class (and more generally, any low-degree of the input) fails to detect the planted vector.
  arXiv  Detail & Related papers  (2021-05-31T16:10:49Z)
• Linear Bandits on Uniformly Convex Sets [88.3673525964507]
  Linear bandit algorithms yield $tildemathcalO(nsqrtT)$ pseudo-regret bounds on compact convex action sets. Two types of structural assumptions lead to better pseudo-regret bounds.
  arXiv  Detail & Related papers  (2021-03-10T07:33:03Z)
• Optimal Regret Algorithm for Pseudo-1d Bandit Convex Optimization [51.23789922123412]
  We study online learning with bandit feedback (i.e. learner has access to only zeroth-order oracle) where cost/reward functions admit a "pseudo-1d" structure. We show a lower bound of $min(sqrtdT, T3/4)$ for the regret of any algorithm, where $T$ is the number of rounds. We propose a new algorithm sbcalg that combines randomized online gradient descent with a kernelized exponential weights method to exploit the pseudo-1d structure effectively.
  arXiv  Detail & Related papers  (2021-02-15T08:16:51Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.