Intelligent optimization: Literature review and state-of-the-art algorithms (1965–2022)

A Mohammadi, F Sheikholeslam - Engineering Applications of Artificial …, 2023 - Elsevier
Today, intelligent optimization has become a science that few researchers have not used in
dealing with problems in their field. Diversity and flexibility have made the use, efficiency …

Nature-inspired metaheuristic search algorithms for optimizing benchmark problems: inclined planes system optimization to state-of-the-art methods

A Mohammadi, F Sheikholeslam, S Mirjalili - Archives of Computational …, 2023 - Springer
In the literature, different types of inclined planes system optimization (IPO) algorithms have
been proposed and evaluated in various applications. Due to the large number of variants …

Quasi-Newton methods for machine learning: forget the past, just sample

AS Berahas, M Jahani, P Richtárik… - … Methods and Software, 2022 - Taylor & Francis
We present two sampled quasi-Newton methods (sampled LBFGS and sampled LSR1) for
solving empirical risk minimization problems that arise in machine learning. Contrary to the …

Inclined planes system optimization: theory, literature review, and state-of-the-art versions for IIR system identification

A Mohammadi, F Sheikholeslam, S Mirjalili - Expert Systems with …, 2022 - Elsevier
Abstract The Inclined Planes System Optimization (IPO) algorithm is recent algorithm that
uses Newton's second law to perform optimization. After conducting a thorough literature …

Doubly adaptive scaled algorithm for machine learning using second-order information

M Jahani, S Rusakov, Z Shi, P Richtárik… - arXiv preprint arXiv …, 2021 - arxiv.org
We present a novel adaptive optimization algorithm for large-scale machine learning
problems. Equipped with a low-cost estimate of local curvature and Lipschitz smoothness …

An overview of stochastic quasi-Newton methods for large-scale machine learning

TD Guo, Y Liu, CY Han - Journal of the Operations Research Society of …, 2023 - Springer
Numerous intriguing optimization problems arise as a result of the advancement of machine
learning. The stochastic first-order method is the predominant choice for those problems due …

Flecs: A federated learning second-order framework via compression and sketching

A Agafonov, D Kamzolov, R Tappenden… - arXiv preprint arXiv …, 2022 - arxiv.org
Inspired by the recent work FedNL (Safaryan et al, FedNL: Making Newton-Type Methods
Applicable to Federated Learning), we propose a new communication efficient second-order …

LSOS: Line-search Second-Order Stochastic optimization methods for nonconvex finite sums

D Di Serafino, N Krejić, N Krklec Jerinkić… - Mathematics of …, 2023 - ams.org
We develop a line-search second-order algorithmic framework for minimizing finite sums.
We do not make any convexity assumptions, but require the terms of the sum to be …

Stochastic gradient methods with preconditioned updates

A Sadiev, A Beznosikov, AJ Almansoori… - Journal of Optimization …, 2024 - Springer
This work considers the non-convex finite-sum minimization problem. There are several
algorithms for such problems, but existing methods often work poorly when the problem is …

Generalization of Quasi-Newton methods: application to robust symmetric multisecant updates

D Scieur, L Liu, T Pumir… - … Conference on Artificial …, 2021 - proceedings.mlr.press
Quasi-Newton (qN) techniques approximate the Newton step by estimating the Hessian
using the so-called secant equations. Some of these methods compute the Hessian using …