ISRES+: an improved evolutionary strategy for function minimization to estimate the free parameters of systems biology models

Bioinformatics, Год журнала: 2023, Номер 39(7)

Опубликована: Июнь 24, 2023

Mathematical models in systems biology help generate hypotheses, guide experimental design, and infer the dynamics of gene regulatory networks. These are characterized by phenomenological or mechanistic parameters, which typically hard to measure. Therefore, efficient parameter estimation is central model development. Global optimization techniques, such as evolutionary algorithms (EAs), applied estimate parameters inverse modeling, i.e. calibrating minimizing a function that evaluates measure error between predictions data. EAs "fittest individuals" generating large population individuals using strategies like recombination mutation over multiple "generations." Typically, only few from each generation used create new next generation. Improved Evolutionary Strategy Stochastic Ranking (ISRES), proposed Runnarson Yao, one EA widely parameters. ISRES uses information at most pair any minimize error. In this article, we propose an strategy, ISRES+, builds on combining all across generations develop better understanding fitness landscape.

Язык: Английский

---

Batch metaheuristic: a migration-free framework for metaheuristic algorithms

Evolutionary Intelligence, Год журнала: 2023, Номер 17(3), С. 1855 - 1887

Опубликована: Авг. 7, 2023

Язык: Английский

---

Lightweight convolutional neural network architecture design for music genre classification using evolutionary stochastic hyperparameter selection

Expert Systems, Год журнала: 2023, Номер 40(6)

Опубликована: Янв. 24, 2023

Abstract Convolutional neural networks (CNNs) have succeeded in various domains, including music information retrieval (MIR). Music genre classification (MGC) is one such task the MIR that has gained attention over years because of massive increase online content. Accurate indexing and automatic these large volumes content require high computational resources, which pose a significant challenge to building lightweight system. CNNs are popular deep learning‐based choice for systems MGC. However, finding an optimal CNN architecture MGC requires domain knowledge both design music. We present MGA‐CNN, genetic algorithm‐based approach with novel stochastic hyperparameter selection CNN‐based task. The proposed unique automating MGA‐CNN evaluated on three widely used datasets compared seven peer rivals, include approaches four manually designed architectures. experimental results show surpasses terms accuracy, parameter numbers, execution time. architectures generated by also achieve accuracy comparable while spending fewer computing resources.

Язык: Английский

---

Genetic model-based success probability prediction of quantum software development projects

Information and Software Technology, Год журнала: 2023, Номер 165, С. 107352 - 107352

Опубликована: Окт. 26, 2023

Quantum computing (QC) holds the potential to revolutionize by solving complex problems exponentially faster than classical computers, transforming fields such as cryptography, optimization, and scientific simulations. To unlock benefits of QC, quantum software development (QSD) enables harnessing its power, further driving innovation across diverse domains. ensure successful QSD projects, it is crucial concentrate on key variables. This study aims identify variables in develop a model for predicting success probability projects. We identified from existing literature achieve these objectives collected expert insights using survey instrument. then analyzed an optimization model, i.e., Genetic Algorithm (GA), with two different prediction methods Naïve Bayes Classifier (NBC) Logistic Regression (LR). The results models indicate that process matures, project significantly increases, costs are notably reduced. Furthermore, best fitness rankings each variable determined NBC LR indicated strong positive correlation (rs=0.945). t-test (t = 0.851, p 0.402>0.05) show no significant differences between calculated (NBC LR). reveal developed based 14 variables, highlights areas where practitioners need focus more order facilitate cost-effective implementation

Язык: Английский

---

Performance prediction model for desalination plants using modified grey wolf optimizer based artificial neural network approach

Desalination and Water Treatment, Год журнала: 2024, Номер 319, С. 100411 - 100411

Опубликована: Май 24, 2024

Desalination represents an effective method for alleviating water scarcity, applying algorithmic techniques to predict the performance of reverse osmosis (RO) desalination plants, Modified Grey Wolf Optimizer (MGWO) based Artificial Neural Networks (ANN) can membrane distillation (MD) equipment. Four experimental inputs are selected: feed salt concentration(35-140 g/h), flow rate(400-600 L/h), evaporator inlet temperature (60-80℃), and condenser (20-30℃). The permeate flux (L/h m2) is selected as output. Ten prediction models were proposed compared with existing (ANN, WOA-ANN, GWO-ANN). results showed that MGWO-ANN model-5 best regression results: R2=99.3%, mean square error (MSE)=0.004. This model outperformed ANN (R2=98.8%, MSE=0.060), WOA-ANN (R2=99.1%, MSE=0.005) GWO-ANN (R2=98.9%, MSE=0.007). Model-5 has a single hidden layer (H=1), 13 nodes (n=13), 10 search agents (SA=10), 75%-20%-05% dataset division. Its residual within acceptable limits (spanning -0.1 0.2). Optimizing number (n) (SA) improve training efficiency accuracy model, capable more accurately predicting plants.

Язык: Английский

---