Article
Biochemical Research Methods
Gemma Massonis, Alejandro F. Villaverde, Julio R. Banga
Summary: MotivationDynamic mechanistic modelling in systems biology has been hindered by complexity and variability, as well as uncertain and sparse experimental measurements. Ensemble modelling has been introduced to mitigate these issues, but is unreliable for predicting non-observable states. In this study, the authors present a strategy to assess and improve the reliability of model ensembles, using a diversity-enforcing technique combined with identifiability and observability analysis. They demonstrate the effectiveness of their approach with models of glucose regulation, cell division, circadian oscillations, and the JAK-STAT signalling pathway.
Article
Operations Research & Management Science
Charles Audet, Sebastien Le Digabel, Renaud Saltet
Summary: This research introduces a new approach to solving blackbox optimization problems by building ensembles of surrogates and quantifying the uncertainty of their predictions, which allows for optimization at a lower computational cost. Computational experiments demonstrate that this method achieves higher precision and lower computational effort compared to traditional stochastic models in solving expensive simulation problems.
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS
(2022)
Article
Energy & Fuels
Alexander Thebelt, Calvin Tsay, Robert M. Lee, Nathan Sudermann-Merx, David Walz, Tom Tranter, Ruth Misener
Summary: This paper proposes a novel data-driven strategy using tree ensembles for constrained multi-objective optimization of black-box problems with heterogeneous variable spaces. It demonstrates competitive performance and sampling efficiency in real-world applications with limited evaluation budgets compared to other state-of-the-art tools.
Article
Computer Science, Artificial Intelligence
Maryam Sabzevari, Gonzalo Martinez-Munoz, Alberto Suarez
Summary: This paper discusses a method of building heterogeneous ensembles by pooling classifiers from different types of homogeneous ensembles. The approach can achieve better prediction results through a suitable combination and significant gains at a fraction of the training cost.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2022)
Article
Chemistry, Medicinal
Francesca Peccati, Sara Alunno-Rufini, Gonzalo Jimenez-Oses
Summary: Thermostability enhancement is crucial for the industrial and biotechnological application of biocatalysts, especially under high temperatures and harsh conditions. Predicting thermostability changes accurately is challenging due to the modifications of free energy of unfolding. In this study, we develop a new computational protocol combining global conformational sampling and energy prediction using AlphaFold and Rosetta, to quantitatively predict thermostability changes during laboratory evolution of acyltransferase LovD and lipase LipA. We demonstrate that ensembles based on AlphaFold models provide more accurate and robust thermostability predictions compared to ensembles based solely on crystallographic structures, which introduce a distortion in computed thermostabilities.
JOURNAL OF CHEMICAL INFORMATION AND MODELING
(2023)
Article
Biochemistry & Molecular Biology
Forrest Hurley, Christine Heitsch
Summary: RNAprofiling 2.0 identifies dominant helices/stems and organizes them into profiles based on suboptimal sampling data. It enhances profile selection and visualizes relationships in a decision tree, allowing a greater understanding of trade-offs among different possible base pairing combinations. This analysis is made accessible to experimental researchers in a portable format as an interactive webpage.
JOURNAL OF MOLECULAR BIOLOGY
(2023)
Article
Engineering, Mechanical
Kai Cheng, Zhenzhou Lu, Sinan Xiao, Xiaobo Zhang, Sergey Oladyshkin, Wolfgang Nowak
Summary: The paper proposes a fully decoupled simulation method for reliability-based design optimization using thermodynamic integration and parallel tempering. By treating design parameters as uniformly distributed random variables and using importance sampling, the method provides robust solutions for various nonlinear constraint problems.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2021)
Article
Mathematics
Jorge Perez-Aracil, Carlos Camacho-Gomez, Eugenio Lorente-Ramos, Cosmin M. Marina, Laura M. Cornejo-Bueno, Sancho Salcedo-Sanz
Summary: This paper proposes new probabilistic and dynamic strategies for creating multi-method ensembles based on the CRO-SL algorithm. Two different probabilistic strategies are analyzed to improve the algorithm. The performances of the proposed ensembles are tested for different optimization problems, comparing the results with existing algorithms in the literature.
Article
Engineering, Aerospace
Xiaohui Wang, Hao Zhang, Shengzhou Bai, Yuxian Yue
Summary: This paper introduces a novel optimization method, the hybrid-resampling particle swarm optimization (HRPSO) algorithm, which offers higher efficiency for constellation design. Simulation results demonstrate that the HRPSO algorithm outperforms traditional methods, making it a practical choice for constellation design.
Article
Optics
Owen Miller, Kyoungweon Park, Richard A. Vaia
Summary: This paper presents the optimal design of colloidal nanoparticle ensembles, focusing on the largest possible optical cross-section and its experimental demonstration. The study combines theory and experiment to derive general bounds on maximum cross-sections and apply an analytical antenna model to illustrate the potential of resonant nanorods. The authors use a modified synthesis approach to produce gold nanorod ensembles with small polydispersity, and propose an extinction metric for predicting polydispersity. The findings have broad applicability to plasmonic materials and provide a roadmap for achieving the largest optical response in nanoparticle ensembles.
Article
Chemistry, Multidisciplinary
Sabina-Adriana Floria, Marius Gavrilescu, Florin Leon, Silvia Curteanu
Summary: This paper utilizes various biologically inspired optimization algorithms to train multilayer perceptron neural networks and generate good regression models. By combining different optimization algorithms into a hybrid ensemble optimizer, the search capability is improved. Experimental results show that the neural networks generated by the hybrid multiple elite strategy are the most dependable regression models.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Luiz C. F. Ribeiro, Gustavo H. de Rosa, Douglas Rodrigues, Joao P. Papa
Summary: This paper proposes creating Convolutional Neural Networks ensembles through Single-Iteration Optimization to address the issue of highly specific hyperparameter settings. The results demonstrate that this method can achieve promising results while reducing the time required.
Article
Computer Science, Artificial Intelligence
Xiaogang Qi, Zhinan Li, Chen Chen, Lifang Liu
Summary: This paper proposes two algorithms for node deployment: an improved virtual force algorithm and a resampling particle swarm optimization algorithm embedded with virtual force. Simulation results show that the improved virtual force algorithm can quickly make the network reach a stable state and achieve high coverage rates, while the resampling particle swarm optimization algorithm embedded with virtual force has the highest coverage rate among multiple algorithms.
APPLIED INTELLIGENCE
(2022)
Article
Biochemical Research Methods
Priyojit Das, Tongye Shen, Rachel Patton McCord
Summary: Chromosome structures vary between different cell types and regions, and are influenced by factors such as epigenetic state.
PLOS COMPUTATIONAL BIOLOGY
(2022)
Article
Computer Science, Hardware & Architecture
Serena Wang, Maya Gupta, Seungil You
Summary: The method proposes a strategy of jointly optimizing the evaluation order of base models and early-stopping thresholds to speed up the evaluation process of classifier ensembles without harming accuracy. Experimental results show that the method outperforms previous fixed orderings across different types of ensembles.
ACM JOURNAL ON EMERGING TECHNOLOGIES IN COMPUTING SYSTEMS
(2021)
Article
Biochemistry & Molecular Biology
Xuankun Zeng, Arzu Uyar, Dexin Sui, Nazanin Donyapour, Dianqing Wu, Alex Dickson, Jian Hu
BIOCHEMICAL JOURNAL
(2018)
Article
Multidisciplinary Sciences
Michelle L. Milstein, Breyanna L. Cavanaugh, Nicole M. Roussey, Stefanie Volland, David S. Williams, Andrew F. X. Goldberg
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2020)
Article
Chemistry, Physical
Nicole M. Roussey, Alex Dickson
JOURNAL OF CHEMICAL PHYSICS
(2020)
Article
Chemistry, Multidisciplinary
Nazanin Donyapour, Matthew Hirn, Alex Dickson
Summary: This work examines methods for predicting the partition coefficient of small molecules, using atomic attributes transformed into molecular features and training models with neural networks. The best prediction accuracies are obtained using atomic attributes generated with the CHARMM generalized force field and 2D molecular structures.
JOURNAL OF COMPUTATIONAL CHEMISTRY
(2021)