Article
Computer Science, Artificial Intelligence
Junali Jasmine Jena, Samarendra Chandan Bindu Dash, Suresh Chandra Satapathy
Summary: This paper focuses on the stability analysis of swarm-based optimization algorithms, especially the Social Group Optimization (SGO) algorithm. The results show that the algorithm performs better within a stable range, ensuring convergence.
COMPLEX & INTELLIGENT SYSTEMS
(2022)
Article
Automation & Control Systems
Noud Mooren, Gert Witvoet, Tom Oomen
Summary: This paper aims to develop an adaptive feedforward controller for non-resetting point-to-point motion tasks by using a data-driven feedforward controller. An approximate optimal instrumental variable (IV) estimator with real-time bootstrapping is employed to update the feedforward parameters in a closed-loop setting. A case study on a wafer-stage and experimental validation on a benchmark motion system demonstrate the performance improvement.
INTERNATIONAL JOURNAL OF ROBUST AND NONLINEAR CONTROL
(2023)
Article
Engineering, Multidisciplinary
Shane T. Barratt, Stephen P. Boyd
Summary: The article introduces a powerful proximal gradient method for least squares auto-tuning to find hyper-parameters in least squares problems. Numerical experiments using the MNIST dataset demonstrate the effectiveness of the method, cutting the test error of standard least squares in half.
ENGINEERING OPTIMIZATION
(2021)
Article
Computer Science, Artificial Intelligence
Juyeon Seo, Kyeonghun Kim, Sangmin Seo, Sanghyun Park
Summary: In this study, we propose the DARK tuning system to improve the performance of Redis, an in-memory key-value store. The tuning was performed by classifying knobs related to persistence methods. We also propose the Cross-GA method, which alternates between prediction and alignment using Genetic Algorithm to improve throughput and latency simultaneously. Through performance evaluations using Memtier-benchmark, the proposed method achieved up to 39.8% improvement in throughput and 71.3% improvement in latency compared to the existing configuration.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Yuqi Xiao, Yongjun Wu
Summary: In this study, MOA is applied to visual target tracking for the first time, and a novel meta-heuristic tracking algorithm with efficiency and precision is obtained. By designing the updating mechanism and introducing a chaos algorithm, the drawbacks of the standard MOA algorithm in local extremals, convergence speed, and efficiency are improved. Sufficient tracking experiments in various tracking scenes show that our tracker performs great in terms of efficiency, robustness and accuracy.
IMAGE AND VISION COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Paul Escapil-Inchauspe, Gonzalo A. Ruz
Summary: This paper applies physics-informed neural networks (PINNs) to forward physical problems and optimizes the hyperparameters using Gaussian processes-based Bayesian optimization. The method is applied to Helmholtz equation for bounded domains and thoroughly studied in terms of performance, collocation points density, and frequency, confirming its applicability and necessity.
Article
Computer Science, Hardware & Architecture
Pavel Skrabanek, Natalia Martinkova
Summary: Computer vision systems' performance is influenced by their design and parameter setting, particularly the conversion of RGB images to grayscale format. By appropriately setting the weights and tuning the system, the performance can be improved and sensitivity to changes in data distribution can be minimized.
Article
Mathematics, Applied
Ran Xin, Usman A. Khan, Soummya Kar
Summary: This paper explores the decentralized minimization problem in a directed network and proposes a stochastic first-order gradient method called GT-SARAH. The method is able to find an epsilon-accurate first-order stationary point with a lower gradient complexity and achieves linear speedup in a big-data regime.
SIAM JOURNAL ON OPTIMIZATION
(2022)
Article
Computer Science, Information Systems
Donghua Chen, Runtong Zhang, Robin Guanghua Qiu
Summary: This article discusses the need for a self-tuning system to improve MapReduce performance in a complicated Hadoop environment. The proposed Catla system utilizes succinct templates and proper schemes of MapReduce algorithms to facilitate tuning and optimization. The comprehensive evaluation of the Catla system includes multiple tuning approaches to identify optimal Hadoop parameters for deployed MapReduce jobs.
IEEE SYSTEMS JOURNAL
(2021)
Article
Computer Science, Information Systems
Hui Dou, Kang Wang, Yiwen Zhang, Pengfei Chen
Summary: To achieve better performance, big data processing frameworks usually have a large number of performance-critical configuration parameters. Manually configuring these parameters is time-consuming, so there is a need for automatic tuning. This paper proposes a black-box approach, ATConf, to automatically tune the configuration parameters for BDPFs. Experimental results show that ATConf can reduce the execution time by 46.52% compared to the default configuration.
CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Rosita Guido, Maria Carmela Groccia, Domenico Conforti
Summary: Hyperparameter tuning is essential for improving model performance in machine learning. This research focuses on classifying imbalanced data using cost-sensitive support vector machines and proposes a multi-objective approach to optimize the model's hyperparameters. The algorithm is presented in a basic version and an improved version utilizing genetic algorithms and decision trees. Experimental results demonstrate the importance of using appropriate evaluation measures for assessing the classification performance of imbalanced data classification models.
Article
Computer Science, Theory & Methods
Jeremie Coullon, Leah South, Christopher Nemeth
Summary: Stochastic gradient Markov chain Monte Carlo (SGMCMC) is a popular algorithm for scalable Bayesian inference. However, tuning hyperparameters such as step size or batch size is currently a manual and non-principled process. We propose a bandit-based algorithm that minimizes Stein discrepancy to automatically tune the SGMCMC hyperparameters.
STATISTICS AND COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Ying Liu, Gongfa Li, Du Jiang, Juntong Yun, Li Huang, Yuanmin Xie, Guozhang Jiang, Jianyi Kong, Bo Tao, Chunlong Zou, Zifan Fang
Summary: In this study, a dynamic ensemble multi-strategy bald eagle search (DMBES) algorithm is proposed to address the problems of BES, such as local optimums, limited diversity, and slow convergence. By constructing a nonlinear control factor, introducing a Levy flight strategy, and proposing a dynamic selection strategy, DMBES algorithm outperforms other algorithms in the numerical experiments.
APPLIED SOFT COMPUTING
(2023)
Article
Neurosciences
Peiyuan Tian, Guanghua Xu, Chengcheng Han, Xun Zhang, Xiaowei Zheng, Fan Wei, Sicong Zhang, Zhe Zhao
Summary: This paper proposes a new method based on underdamped second-order stochastic resonance to accurately quantify visual fatigue caused by SSVEP paradigm, and compares it with traditional methods. The results show that the quantification value obtained by the new method is closer to the subjective gold standard score, indicating that the new method is more reliable.
FRONTIERS IN NEUROSCIENCE
(2023)
Article
Computer Science, Artificial Intelligence
Jesus-Adolfo Mejia-de-Dios, Efren Mezura-Montes, Marcela Quiroz-Castellanos
Summary: This work proposes an automated parameter tuning problem modeled as a bilevel optimization problem, solved using surrogate models. Experimental results show that the proposed method outperforms the compared approach in finding better configurations with fewer calls to the target algorithm.
APPLIED INTELLIGENCE
(2021)