Online (multinomial) logistic bandit: Improved regret and constant computation cost
YJ Zhang, M Sugiyama - Advances in Neural Information …, 2024 - proceedings.neurips.cc
This paper investigates the logistic bandit problem, a variant of the generalized linear bandit
model that utilizes a logistic model to depict the feedback from an action. While most existing …
model that utilizes a logistic model to depict the feedback from an action. While most existing …
Second-order optimization with lazy hessians
We analyze Newton's method with lazy Hessian updates for solving general possibly non-
convex optimization problems. We propose to reuse a previously seen Hessian for several …
convex optimization problems. We propose to reuse a previously seen Hessian for several …
Greedy quasi-Newton methods with explicit superlinear convergence
A Rodomanov, Y Nesterov - SIAM Journal on Optimization, 2021 - SIAM
In this paper, we study greedy variants of quasi-Newton methods. They are based on the
updating formulas from a certain subclass of the Broyden family. In particular, this subclass …
updating formulas from a certain subclass of the Broyden family. In particular, this subclass …
Safe grid search with optimal complexity
Popular machine learning estimators involve regularization parameters that can be
challenging to tune, and standard strategies rely on grid search for this task. In this paper …
challenging to tune, and standard strategies rely on grid search for this task. In this paper …
Differentially private inference via noisy optimization
Differentially private inference via noisy optimization Page 1 The Annals of Statistics 2023,
Vol. 51, No. 5, 2067–2092 https://doi.org/10.1214/23-AOS2321 © Institute of Mathematical …
Vol. 51, No. 5, 2067–2092 https://doi.org/10.1214/23-AOS2321 © Institute of Mathematical …
Finite-sample analysis of -estimators using self-concordance
DM Ostrovskii, F Bach - 2021 - projecteuclid.org
The classical asymptotic theory for parametric M-estimators guarantees that, in the limit of
infinite sample size, the excess risk has a chi-square type distribution, even in the …
infinite sample size, the excess risk has a chi-square type distribution, even in the …
Exploration via linearly perturbed loss minimisation
We introduce\emph {exploration via linear loss perturbations}(EVILL), a randomised
exploration method for structured stochastic bandit problems that works by solving for the …
exploration method for structured stochastic bandit problems that works by solving for the …
Improved Regret Bounds of (Multinomial) Logistic Bandits via Regret-to-Confidence-Set Conversion
Logistic bandit is a ubiquitous framework of modeling users' choices, eg, click vs. no click for
advertisement recommender system. We observe that the prior works overlook or neglect …
advertisement recommender system. We observe that the prior works overlook or neglect …
Global linear convergence of Newton's method without strong-convexity or Lipschitz gradients
We show that Newton's method converges globally at a linear rate for objective functions
whose Hessians are stable. This class of problems includes many functions which are not …
whose Hessians are stable. This class of problems includes many functions which are not …
Semi-discrete optimal transport: Hardness, regularization and numerical solution
B Taşkesen, S Shafieezadeh-Abadeh… - Mathematical Programming, 2023 - Springer
Semi-discrete optimal transport problems, which evaluate the Wasserstein distance between
a discrete and a generic (possibly non-discrete) probability measure, are believed to be …
a discrete and a generic (possibly non-discrete) probability measure, are believed to be …