Benchmarking the performance of Bayesian optimization across multiple experimental materials science domains

npj Computational Materials, Год журнала: 2021, Номер 7(1)

Опубликована: Ноя. 18, 2021

Abstract Bayesian optimization (BO) has been leveraged for guiding autonomous and high-throughput experiments in materials science. However, few have evaluated the efficiency of BO across a broad range experimental domains. In this work, we quantify performance with collection surrogate model acquisition function pairs five diverse systems. By defining acceleration enhancement metrics objectives, find that models such as Gaussian Process (GP) anisotropic kernels Random Forest (RF) comparable BO, both outperform commonly used GP isotropic kernels. demonstrated most robustness, yet RF is close alternative warrants more consideration because it free from distribution assumptions, smaller time complexity, requires less effort initial hyperparameter selection. We also raise awareness about benefits using future campaigns.

Язык: Английский

---

Metamodel-based simulation optimization: A systematic literature review

Simulation Modelling Practice and Theory, Год журнала: 2021, Номер 114, С. 102403 - 102403

Опубликована: Сен. 7, 2021

Язык: Английский

---

Assessment of human health risk from potentially toxic elements and predicting groundwater contamination using machine learning approaches

Journal of Contaminant Hydrology, Год журнала: 2024, Номер 261, С. 104307 - 104307

Опубликована: Янв. 21, 2024

The Rooppur Nuclear Power Plant (RNPP) at Ishwardi, Bangladesh is planning to go into operation within 2024 and therefore, adjacent areas of RNPP gaining adequate attention from the scientific community for environmental monitoring purposes especially water resources management. However, there a substantial lack literature as well datasets earlier years since very little was done beginning RNPP's construction phase. Therefore, this study conducted assess potential toxic elements (PTEs) contamination in groundwater its associated health risk residents part during year 2014–2015. For achieving aim study, samples were collected seasonally (dry wet season) nine sampling sites afterwards analyzed quality indicators such temperature (Temp.), pH, electrical conductivity (EC), total dissolved solid (TDS), hardness (TH) PTEs including Iron (Fe), Manganese (Mn), Copper (Cu), Lead (Pb), Chromium (Cr), Cadmium (Cd) Arsenic (As). This adopted newly developed Root Mean Square index (RMS-WQI) model scenario whereas human assessment utilized quantify toxicity PTEs. In most sites, concentration found higher season than dry Fe, Mn, Cd As exceeded guideline limit drinking water. RMS score mostly classified terms "Fair" condition. non-carcinogenic risks (expressed Hazard Index-HI) revealed that around 44% 89% adults 67% 100% children threshold set by USEPA (HI > 1) possessed through oral pathway season, respectively. Furthermore, calculated cumulative HI throughout period. carcinogenic (CR) PTEs, magnitude decreased following pattern Cr Cd. Although current based on old dataset, findings might serve baseline reduce future hazardous impact power plant.

Язык: Английский

---

Bayesian Optimization is Superior to Random Search for Machine Learning Hyperparameter Tuning: Analysis of the Black-Box Optimization Challenge 2020

arXiv (Cornell University), Год журнала: 2021, Номер unknown

Опубликована: Янв. 1, 2021

This paper presents the results and insights from black-box optimization (BBO) challenge at NeurIPS 2020 which ran July-October, 2020. The emphasized importance of evaluating derivative-free optimizers for tuning hyperparameters machine learning models. was first with a emphasis. It based on (validation set) performance standard models real datasets. competition has widespread impact as (e.g., Bayesian optimization) is relevant hyperparameter in almost every project well many applications outside learning. final leaderboard determined using held-out (hidden) objective functions, where without human intervention. Baselines were set default settings several open-source packages random search.

Язык: Английский

---

Automated Discovery of Trade-Off Between Utility, Privacy and Fairness in Machine Learning Models

Communications in computer and information science, Год журнала: 2025, Номер unknown, С. 127 - 144

Опубликована: Янв. 1, 2025

Язык: Английский

---

Bayesian Optimization of Booster Disinfection Scheduling in Water Distribution Networks

Water Research, Год журнала: 2023, Номер 242, С. 120117 - 120117

Опубликована: Май 23, 2023

Язык: Английский

---

Multi-cavitation states identification of a sewage pump using CEEMDAN and BOA-SVM

Journal of Water Process Engineering, Год журнала: 2024, Номер 61, С. 105299 - 105299

Опубликована: Апрель 17, 2024

Язык: Английский

---

An explainable boosting-based ensemble machine learning model for predicting the properties of date palm fiber reinforced concrete

Sustainable Chemistry and Pharmacy, Год журнала: 2025, Номер 44, С. 101949 - 101949

Опубликована: Фев. 15, 2025

Язык: Английский

---

Towards Development of an Optimization Model to Identify Contamination Source in a Water Distribution Network

Water, Год журнала: 2018, Номер 10(5), С. 579 - 579

Опубликована: Апрель 29, 2018

Protection of the water system is paramount due to negative consequences contaminated on public health. Water resources are one critical infrastructures that must be preserved from deliberate and accidental attacks. qualities examined at treatment plant. However, its quality can substantially during transportation plant consumers’ taps. Contamination in distribution networks (WDNs) a danger have severe health as well an economic social instability. immensely susceptible or attacks complex nature system. Hence, contamination source identification (CSI) topical issue systems require immediate attention researchers order protect mankind adverse effect consuming water. Usually, contaminant event detected by monitoring sensors warning (CWS) installed network. Nevertheless, how derive collected information difficult task tackled evaluate spread for remedial strategies. In past two decades, considerable efforts advancement been made applying various techniques locate WDNs. Each has certain limitations applicability reported literature. This paper presents comprehensive review existing with emphasis their importance technical challenges. Despite series investigations this domain, field yet unified. open research areas still available explore. Consequently, improvement necessary hereby suggested. More importantly, practical application these offer major gap addressed.

Язык: Английский

---

Sequential model based optimization of partially defined functions under unknown constraints

Journal of Global Optimization, Год журнала: 2019, Номер 79(2), С. 281 - 303

Опубликована: Дек. 2, 2019

Abstract This paper presents a sequential model based optimization framework for optimizing black-box, multi-extremal and expensive objective function, which is also partially defined, that it undefined outside the feasible region. Furthermore, constraints defining region within search space are unknown. The approach proposed in this paper, namely SVM-CBO, organized two consecutive phases, first uses Support Vector Machine classifier to approximate boundary of unknown region, second Bayesian Optimization find globally optimal solution In phase next point evaluate chosen by dealing with trade-off between improving current estimate discovering possible disconnected sub-regions. phase, selected as minimizer Lower Confidence Bound acquisition function but constrained main comparison process fixed penalty value infeasible evaluations, under limited budget (i.e., maximum number evaluations). Results related five 2D test functions from literature 80 functions, increasing dimensionality complexity, generated through Emmental-type GKLS software. SVM-CBO proved be significantly more effective well computationally efficient.

Язык: Английский

---