检索结果-内蒙古大学图书馆

An Enhancement of Cuckoo Search Algorithm for Optimal Earthquake Evacuation Space Allocation in Intramuros, Manila City

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Villanueva, Marcus Andre Ching, Charles Matthew Mata, Khatalyn Department of Computer Science Pamantasan ng Lungsod ng Maynila Manila Philippines

The Cuckoo Search Algorithm (CSA), while effective in solving complex optimization problems, faces limitations in random population initialization and reliance on fixed parameters. Random initialization of the population often results in clustered solutions, resulting in uneven exploration of the search space and hindering effective global optimization. Furthermore, the use of fixed values for discovery rate and step size creates a trade-off between solution accuracy and convergence speed. To address these limitations, an Enhanced Cuckoo Search Algorithm (ECSA) is proposed. This algorithm utilizes the Sobol Sequence to generate a more uniformly distributed initial population and incorporates Cosine Annealing with Warm Restarts to dynamically adjust the parameters. The performance of the algorithms was evaluated on 13 benchmark functions (7 unimodal, 6 multimodal). Statistical analyses were conducted to determine the significance and consistency of the results. The ECSA outperforms the CSA in 11 out of 13 benchmark functions with a mean fitness improvement of 30% across all functions, achieving 35% for unimodal functions and 24% for multimodal functions. The enhanced algorithm demonstrated increased convergence efficiency, indicating its superiority to the CSA in solving a variety of optimization problems. The ECSA is subsequently applied to optimize earthquake evacuation space allocation in Intramuros, Manila. Copyright © 2025, The Authors. All rights reserved.

关键词： optimization algorithms

On The Concurrence of Layer-wise Preconditioning Methods and Provable Feature Learning

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhang, Thomas T. Moniri, Behrad Nagwekar, Ansh Rahman, Faraz Xue, Anton Hassani, Hamed Matni, Nikolai University of Pennsylvania United States

Layer-wise preconditioning methods are a family of memory-efficient optimization algorithms that introduce preconditioners per axis of each layer’s weight tensors. These methods have seen a recent resurgence, demonstrating impressive performance relative to entry-wise ("diagonal") preconditioning methods such as Adam(W) on a wide range of neural network optimization tasks. Complementary to their practical performance, we demonstrate that layer-wise preconditioning methods are provably necessary from a statistical perspective. To showcase this, we consider two prototypical models, linear representation learning and single-index learning, which are widely used to study how typical algorithms efficiently learn useful features to enable generalization. In these problems, we show SGD is a suboptimal feature learner when extending beyond ideal isotropic inputs x ∼ N(0, I) and well-conditioned settings typically assumed in prior work. We demonstrate theoretically and numerically that this suboptimality is fundamental, and that layer-wise preconditioning emerges naturally as the solution. We further show that standard tools like Adam preconditioning and batch-norm only mildly mitigate these issues, supporting the unique benefits of layer-wise preconditioning. © 2025, CC BY.

关键词： optimization algorithms

Enhancing PPO with Trajectory-Aware Hybrid Policies

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Liu, Qisai Jiang, Zhanhong Yang, Hsin-Jung Khosravi, Mahsa Waite, Joshua R. Sarkar, Soumik Department of Mechanical Engineering Iowa State University AmesIA50011 United States Department of Computer Science Iowa State University AmesIA50011 United States Department of Industrial and Manufacturing Systems Engineering Iowa State University AmesIA50011 United States Translational AI Center Iowa State University AmesIA50011 United States

Proximal policy optimization (PPO) is one of the most popular state-of-the-art on-policy algorithms that has become a standard baseline in modern reinforcement learning with applications in numerous fields. Though it delivers stable performance with theoretical policy improvement guarantees, high variance, and high sample complexity still remain critical challenges in on-policy algorithms. To alleviate these issues, we propose Hybrid-Policy Proximal Policy optimization (HP3O), which utilizes a trajectory replay buffer to make efficient use of trajectories generated by recent policies. Particularly, the buffer applies the "first in, first out" (FIFO) strategy so as to keep only the recent trajectories to attenuate the data distribution drift. A batch consisting of the trajectory with the best return and other randomly sampled ones from the buffer is used for updating the policy networks. The strategy helps the agent to improve its capability on top of the most recent best performance and in turn reduce variance empirically. We theoretically construct the policy improvement guarantees for the proposed algorithm. HP3O is validated and compared against several baseline algorithms using multiple continuous control environments. Our code is available here. © 2025, CC BY.

关键词： optimization algorithms

Forward-backward splitting under the light of generalized convexity

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Oikonomidis, Konstantinos Laude, Emanuel Patrinos, Panagiotis Kasteelpark Arenberg 10 Leuven3001 Belgium Proxima Fusion Flößergasse 2 Munich81369 Germany

In this paper we present a unifying framework for continuous optimization methods grounded in the concept of generalized convexity. Utilizing the powerful theory of Φ-convexity, we propose a conceptual algorithm that extends the classical difference-of-convex method, encompassing a broad spectrum of optimization algorithms. Relying exclusively on the tools of generalized convexity we develop a gap function analysis that strictly characterizes the decrease of the function values, leading to simplified and unified convergence results. As an outcome of this analysis, we naturally obtain a generalized PL inequality which ensures q-linear convergence rates of the proposed method, incorporating various well-established conditions from the existing literature. Moreover we propose a Φ-Bregman proximal point interpretation of the scheme that allows us to capture conditions that lead to sublinear rates under convexity. © 2025, CC BY.

关键词： optimization algorithms

A Concise Lyapunov Analysis of Nesterov’s Accelerated Gradient Method

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Liu, Jun Department of Applied Mathematics University of Waterloo WaterlooONN2L 3G1 Canada

Convergence analysis of Nesterov’s accelerated gradient method has attracted significant attention over the past decades. While extensive work has explored its theoretical properties and elucidated the intuition behind its acceleration, a simple and direct proof of its convergence rates is still lacking. We provide a concise Lyapunov analysis of the convergence rates of Nesterov’s accelerated gradient method for both general convex and strongly convex functions. Copyright © 2025, The Authors. All rights reserved.

关键词： optimization algorithms

A New and Robust ABC-K-Means Hybrid Approach for Improved Image Clustering

学校读者我要写书评

暂无评论

SSRN

SSRN 2025年

作者： El Haouari, Oussama Mourad, Hana Khrissi, Lahbib El Akkad, Nabil LASET Laboratory of Applied Sciences and Emerging Technologies National School of Applied Sciences of Fez Sidi Mohamed Ben Abdellah University Fez Morocco LIPI Laboratory of Interdisciplinary Computer Science and Physics Normal Superior School of Fez Sidi Mohamed Ben Abdellah University Fez Morocco

Clustering remains a critical task in image analysis, yet traditional K-means methods frequently suffer from local optima issues, leading to suboptimal clustering, particularly in complex datasets. In this study, we present a novel hybrid approach that integrates the Artificial Bee Colony (ABC) algorithm with K-means to optimize initial centroid placement and improve clustering accuracy. By leveraging ABC's global optimization capabilities, our approach enhances K-means’ robustness and convergence, providing more accurate clustering results. Comparative experiments demonstrate that our ABC-K-means method consistently outperforms other K-means-based hybrid algorithms, including those incorporating Particle Swarm optimization (PSO), Simulated Annealing (SA), and Genetic algorithms (GA). This enhanced method addresses critical limitations in traditional clustering algorithms, establishing a promising solution for high-quality image clustering tasks. © 2025, The Authors. All rights reserved.

关键词： optimization algorithms

A Linear Model for Irrigation Canals Operating in Real Time Applied in ASCE Test Cases

学校读者我要写书评

暂无评论

Water (Switzerland) 2025年第9期17卷 1368-1368页

作者： Bonet, Enrique Yubero, Maria Teresa Bascompta, Marc Alfonso, Pura Les Bases Avenue 61-73 Manresa08242 Spain

In the context of irrigation canal flow, numerical models developed to accurately estimate canal behavior based on gate trajectories are often highly complex. Consequently, hardware limitations make it significantly more challenging to implement these models locally at gate devices. In this regard, one of the most significant contributions of this paper is the concept of the hydraulic influence matrix (HIM) and its application as a linear model to estimate the water surface flow in irrigation canals, integrated within an irrigation canal controller to facilitate real-time operations. The HIM model provides a significant advantage by quickly and accurately computing water level and velocity perturbations in open-flow canals. This capability empowers watermasters to apply this linear free-surface model in both unsteady and steady flow conditions, enabling real-time applications in control algorithms. The HIM model was validated by comparing water-level estimates under various perturbations with results from software using the full Saint-Venant equations. The test involved introducing a 10% perturbation in gate movement over a specified time period in two different test cases, resulting in a flow discharge increase of more than 10% in each test case. The results showed maximum absolute errors below 7 cm and 0.2 cm, relative errors of 0.7% and 0.023%, root mean square errors ranging from 2.4 to 0.07 cm, and Nash–Sutcliffe efficiency values of approximately 0.95 in the first and second test cases, respectively, when compared to the full Saint-Venant equations. This highlights the high precision of the HIM model, even when subjected to significant disturbances. However, larger gate movement disturbances (exceeding 10%) should be planned in advance rather than managed in real time. © 2025 by the authors.

关键词： optimization algorithms

A new tail bound for the sum of bounded independent random variables

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Loper, Jackson Regier, Jeffrey Department of Statistics University of Michigan United States

We construct a new tail bound for the sum of independent random variables for situations in which the expected value of the sum is known and each random variable lies within a specified interval, which may be different for each variable. This new bound can be computed by solving a two-dimensional convex optimization problem. Simulations demonstrate that the new bound is often substantially tighter than Hoeffding’s inequality for cases in which both bounds are applicable. Copyright © 2025, The Authors. All rights reserved.

关键词： optimization algorithms

Optimal Criteria for Best Subset Selection

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhu, Zhihan Zhang, Yanhao Xia, Yong The School of Mathematical Sciences Beihang University Beijing100191 China

This paper introduces two novel criteria: one for feature selection and another for feature elimination in the context of best subset selection, which is a benchmark problem in statistics and machine learning. From the perspective of optimization, we revisit the classical selection and elimination criteria in traditional best subset selection algorithms, revealing that these classical criteria capture only partial variations of the objective function after the entry or exit of features. By formulating and solving optimization subproblems for feature entry and exit exactly, new selection and elimination criteria are proposed, proved as the optimal decisions for the current entry-and-exit process compared to classical criteria. Replacing the classical selection and elimination criteria with the proposed ones generates a series of enhanced best subset selection algorithms. These generated algorithms not only preserve the theoretical properties of the original algorithms but also achieve significant meta-gains without increasing computational cost across various scenarios and evaluation metrics on multiple tasks such as compressed sensing and sparse regression. © 2025, CC BY.

关键词： optimization algorithms